Ask HN: Tell us about your project that's not done yet but you want feedback on | Hacker News Reader

Ask HN: Tell us about your project that's not done yet but you want feedback on

195 pointsabj457 comments

A lot of times with side projects I wished I had gotten feedback early on, before I spent a lot of time on an inefficient direction. I wonder if people wait too long to publish something before it is fully polished, then realized that the polishing wasn't needed.

I'm interested to see things that people would have never published otherwise. I know a lot of my projects never make it to a published phase, but I still would have been interested in knowing the general reception. Please drop your projects here!

View on HN

I'm writing "The Impostor's Guide To The Shell". My theory is that reading popular open-source codebases, and Googling stuff you don't understand until you grok it, is a great way to overcome impostor's syndrome. The goal is to take a reader from thinking "Maybe a 10x engineer could understand that library, but I never could..." to "Oh, that's all it does?".

The guide covers things like:

- shebangs

- exit codes

- parameter expansion

- file permissions

- how to look up docs via "man" and "help"

- And a lot more.

The codebase I'm starting with is a Ruby version manager (written in bash) called RBENV. I've published it onto a platform called HelpThisBook.com, a platform to help authors get feedback from early readers (co-created by Rob Fitzpatrick, author of "The Mom Test", "Write Useful Books", etc.). Instructions on how to leave feedback should be given when you open the link below.

https://helpthisbook.com/richie/impostors-guide-to-the-shell

Too late to edit my original post, so adding this comment to say that the above link just covers RBENV’s shim file code, not the whole codebase.

I have a 600+ page Google doc which covers the remaining code, and which I mostly wrote for my own edification. I was worried about overwhelming my potential audience by showing it all to them at once, and I want to get feedback on just this first section, before releasing the rest.

We’ll see how it goes with just this small first section.

If it helps anyone reading this, everything in tech boils down to "Oh, that's all it does?".

This book looks great - I love the style and I am trying to improve my shell skills.

Is there anywhere I can pre-order it?

Oh wow, thanks for the compliment! You're the first person to ask haha.

The first step in this process is making sure I'm even writing something that's useful to my target audience. So for now, I'm laser-focused on getting reader feedback via that HelpThisBook.com site.

That said, I'm continuing to add more free content to that site on a weekly basis. I recommend watching that space for more updates.

Thanks for reaching out!

Wow, what a great idea for a thread!

I'm trying to pare down my personal projects to just the really exciting ones, so I don't have much, butni think the most appropriate to the thread is https://github.com/EternityForest/iot_devices

It's mean to be a cross-framework library for creating device integrations, so you can, say, write a handler for RTL SDR weather stations, and use it in a simple script up to a mega framework.

I kind of dislike the way HASS and others handle automations where they have special purpose primitives for everything that needs lots of hand written code.

I just have config entries, they must be strings, and data points, they can be strings, numbers, bytes, or objects. You can put metadata on them. There's also a few other utilities like the ability to make subdevices, and the ability to request things from the host.

There are no special subclasses, a light bulb is just a device with a brightness point.

It currently runs my security system with object detection recording, QR decoding if desired, multiple regions, motion detection without decoding every frame, and subsecond latency streaming to the browser, a nice recordings browser that can view a recording while it's being made, etc.

https://pastmaps.com

Been working on this as a new way to find, explore, and view old historical maps and aerials of my area. Still heavily in development but have been surprised by some early traffic stats (1-2K organic uniques / mo and growing 200%+ m/m right now)

Hoping to add in more advanced map tooling within the next week or 2, including new basemap options, 3d terrain view, and then a proper search box which I've been pushing off for far too long

Have you thought of running super resolution on the images?

I actually think that's a great idea! I considered doing it as part of my ingestion pipeline when starting out but honestly got a bit overwhelmed with all the options so put it on my backburner. Maybe I should take another look. I think it would be an awesome way to enable even more crisp zooming

That's really cool! Just wanted to share the following site as the city of Toronto created something similar a few months ago https://map.toronto.ca/torontomaps/

Amazing, just bookmarked this!

Seems like it's the same app but Calgary has this too. The maps around the time of our big flood in 2013 are pretty interesting.

Obviously I want to buy a high quality print of the map focused on an area.

Damn, for a second I thought this was like weedmaps but for linguini (which would be an awesome site) and then I looked a bit closer. Still cool though!

This is really cool, makes me nostalgic for a lot of time georeferencing random maps from the library of congress or from municipal archives etc.

In that same vein, you could add a feature where users could contribute georeferenced map files for community review and approval--I think that would really increase your scalability. I see you have an email for that currently.

I love love love this direction. And yea, I have a basic email but I really think a super simple upload and maybe even dead-easy georeference tool for folks to contribute data is a no-brainer! I'm bumping this up on my backlog, thank you!

Amazing, can't wait to see where you end up! Let me know if you need a hand too. I love web mapping projects--I got into programming through QGIS/leaflet on the way to a couple degrees in GIS and urban planning.

Your site is similar to https://www.oldmapsonline.org/

I have found the ui for map difficult to use in the past.

Yea I used to use this site quite a bit but it is clunky as all hell. I found myself cobbling together my own tools or stitching 3-5 different resources together over time for my research and this eventually just led to me starting work on Pastmaps. I'm hoping I can build something far more comprehensive, modern, and intuitive

I love it! A suggestion for much further down the line: a timeline on the map which composites many maps from a similar time period so you can see them all stitched together (somewhat like how https://skyvector.com/ stitches together multiple sectional charts into a continuous map, though I know it can't be as seamless). Or you could attempt to run some extra processing steps to warp the map to match the background map's projection. Those are both big undertakings, though.

Very doable, these are called mosaics in the GIS world and it's actually not that crazy to do

Really love this suggestion, I could see it making map exploration far easier since now you just need to explore by year instead of by map and by year (reducing dimensionality by 1!). Thank you, I'm going to chew on this a bit more but I really think there's something here

I found a map of the area around my family’s farm in 1894 super quick. Awesome site!

This is super cool, thank you for making it!

An app for language learning by reading books in the target language. It takes the book (any epub book) in the target language add inserts translations:

—Arthur —dijo con tono cortante, ["Arthur," he said sharply.] y su voz sonó como el chasquido de una ratonera—, [and his voice sounded like the click of a mousetrap.]

All of the existing apps that I checked only display translation upon clicking a word (like in Kindle), but that 1) doesn't work well when you're only starting to learn; 2) takes too much from the reading experience, and becomes a chore. Also, most of them only allow a limited selection of books.

I'm using it myself for over a month now, and enjoy reading with it a lot. I feel almost no friction using it, it's the usual pleasure of reading a book with added exploration of a new language.

The problems:

Translating can be hard or / and expensive. Cheaper / free translations are often incorrect, and struggle with idioms. DeepL would cost > 20€ for a long book. ChatGPT hallucinates, and adds to the original. I'm using Google Translate for now and it's good enough for me, but I don't feel it's good enough to charge for. It often mixes genders (as opposed to ChatGPT which can deduct them from context I guess), and occasionally mistranslates.

Would you want to use such app even with the often erroneous automated translations?

That sounds cool, I had a similar idea a while ago but it was a browser extension. I only pulled out keywords from news articles with spacy and then ran them through deepl, I was trying to make automated flashcards to help learn to read articles in your target language.

Have you tried llama2? Running that yourself might be cheaper, you could also maybe crowd source translation fixes eventually somehow.

This sounds very similar to the books manually curated using Ilya Frank's Reading Method: http://english.franklang.ru/index.php/ilya-frank-s-reading-m...

If it works well I would use it (and pay for it)

I'm interested in something like this. I would be even more interested in a curated list of books by language level. Maybe you can offer a couple of books per level and cache those translations, minimizing costs and giving users an easy way to get started reading in a new/intermediate language.

This is very cool, looking forward to it! I've been doing the same thing with Spanish Wikipedia articles for a while, using a few lines of Bash + Regex. I was using Apertium for it. https://apertium.org/ It's definitely worse than most ML-based solutions, but it works reliably, deterministic, and fast; you can run it entirely offline. With Spanish translations, the main problem I was facing is lack of vocabulary, so I created https://github.com/phil294/apertium-eng-spa-wiktionary which about doubles the amount of recognized words, albeit with wonky grammar.

€20 is not expensive to translate a book. Google translate is of such low quality that it is not suitable for any real world use like yours.

Can you let the user herself pay the translation cost when loading her favorite book into your software?

Yeah I intend to offer different translation options and be transparent with pricing

https://github.com/ambrosiogabe/MathAnimation

If you’ve ever tried to make a mathematical animation (think 3Blue1Brown), it’s a real pain. I was using manim for awhile to make animations for my YT channel, but the whole iteration process felt very slow and repetitive. So I thought I would recreate manim over the weekend, except with a GUI and real-time feedback. It’s been a year and a half and I’m hoping this weekend will be done soon so I can move on and start making videos again.

So far, it does a lot, but it still needs a lot of polish and refinement. The readme gives some gifs and a better idea of the feature set right now.

I remember watching the video you made on it[0] and was thoroughly impressed! Small world.

I had an application for this recently but the name made it hard to find :(

[0] https://www.youtube.com/watch?v=iydG-e1dQGA

Ah thanks! And yea, naming things has never been a strength for me. I’ll have to come up with something more memorable once it’s ready for an alpha release :)

AniMath?

Thanks for the recommendation! I’ll have to workshop it a bit more, when I see this I almost read AntiMath which would be an unintended mistake haha.

that looks really good!

Bitgrid[1][2] is a continuation of an idea I've been carrying around since the 1980s, and tweaking from time to time. If I can maintain focus long enough, I intend to learn RTL, VHDL or Verilog well enough to get it entered in one of the Google open source chip shuttles. I think this approach to computing if applied to neural networks may offer a cheap way to get far more compute, perhaps exaflop performance out of a single chip.

It's an FPGA without routing hardware, which on the face of it, is the stupidest thing to do. However, because of the clocked nature of the bitgrid, you all data only has to travel to neighboring cells, so lines will be short, and clock rates should be able to be up in the gigahertz range. Instead of worrying about how quickly you can get a signal from one side of the chip to the other, the latency will be 1/2 of the number of cells across, if the signal takes a straight line.

However... everything along that path can be compute, and routing should be trivial. It avoids the trap with GPUs where they are Turing complete, thus hard to reason about. Like an excel spreadsheet, you can track dependencies, and know exactly where a given bit came from. The chip as a whole, on the other hand, is Turing complete.

If anyone knows how much energy in FemtoJoules an 4 bit latch and a 4 bit input LUT take up, and a static ram cell... that would help in estimating the real world power consumption/feasibility of this thing. I can't find a good answer anywhere.

[1] https://github.com/mikewarot/Bitgrid

[2] https://bitgrid.blogspot.com/2005/03/bitgrid-story.html

Are you sure this can run backprop?

There are many ways you could do backprop... if you want to probe the values straight from the middle of the grid, you could do that. You could poke new values right into the middle of things (by using configuration/random access mode), but everything else is going to pause while doing so. Or you could ripple in new weights along with data.

To me, this feels like starting from where Turing stood and inventing a new type of computer. The sky is the limit.

Have you contrasted this with cellular automata? It seems like an advanced form of that type of construction. Could possibly benefit from some formalization.

Cellular automata apply the same rule across a grid, a bitgrid is an array of lookup tables, essentially stuffed between the cells of a static RAM. The RAM holds the values in the tables, and an latches maintain the state of each cell's inputs. Thus each cell in a bitgrid is individually programmable.

A provision needs to be added to be able to read and or write (override) inputs for debugging or other purposes, such as testing, updating constants, etc.

I agree that formalization is required. I've built an emulator, and hand coded some logic into it as a test, and can simulate a 1024x1024 grid at about 35 Hz on my desktop pc (half that on my laptop).

My near term goals are to be able to take an expression, run it through a set of tools to be written, and then feed it into the simulator, and run it.

I can guess at the energy required to change states, and thus get a rough estimate as to power usage/efficiency/speed, etc. If I get myself to the point where I've got a chip designed, I'm sure the EDA tools can give me far more accurate numbers.

Oh, I meant formalization as in, this is clearly not either a turing machine or a lambda calculus, but as it appears turing-complete, it must be equivalent, and could benefit from formalization in that computer-science/mathematical sense. Then we outsiders would be able to see what we're looking at.

You could consider a BitGrid to be a giant state machine, given any current set of inputs, and the program loaded into it, it is possible to know exactly what the next state will be. The connectivity turns it into a complex system. The number of possible states is somewhere between 1 and 16^cellcount

I'm not sure that it's possible to compute (in a reasonable amount of time, or maybe ever) the exact number of possible states a bitgrid may have, given its current program.

It's my conjecture that finding this exact number is equivalent to solving the halting problem.

---

Also, any computation in a Bitgrid is an acyclic directed graph. Because there is a delay between cells, you can only effect future states.

A little niche, but https://run-in.app is an online encounter manager for DnD 5e that I've been (slowly) building this year. I initially built it because I was dming a one-shot for some friends and got overwhelmed tracking the health and position of 6 enemies + the party on pen and paper.

I wasn't happy with the solutions I found to this online, and thought it'd be fun to try and build one myself.

The main features for creating and running an encounter are ready: 1. Adding any 5e SRD mob or a custom mob to battle maps 2. Dragging mobs around the battle map 3. Turn and health tracking 4. Status effect tracking

But there are a few bugs to iron out, and more features to come!

As a heads up, if you visit the site and get a cold start of the container, it might take a while to be properly interactive. This is, unsurprisingly, the first thing to fix on the roadmap.

Something like this is exactly what I've been looking for! Barebones, easy to use, and snappy. I find most other encounter builders are way to bloated for my purposes. But this seems to strike a really nice middle ground of convenience with all the right features.

I'm so happy to hear that! Please also let me know if you have any feature requests you think would really improve your experience

This looks neat! Have you thought of putting a demo video on the frontpage?

That's actually a great idea and would be much more engaging than a flat image. Cheers :)

I just loaded a page from Europe, it took few seconds to load the images. I thought the page was broken, there were no loader of any kind.

I created ContainerSSH[1] in 2020 which is an SSH server that drops you in a container instead of launching a shell directly. (It's not a containerized SSH, it talks to the Docker/Kubernetes API.) It has since become a CNCF Sandbox project The main use case is creating jump servers and remote dev environments, but it can also be used as a honeypot due to its audit log features that allow you to reconstruct the SSH interaction accurately. We did a few demos where we showed the SSH interactions on a honeypot in real time on a web interface. The authentication and dynamic configuration of containers is done via webhooks.

A now-comaintainer from CERN also contributed X11 and port forwarding support to it and is now continuing the development of the project. Since I switched industries and am nowhere near containers anymore, the project could really use a few contributors.

A few feature ideas we had but never implemented:

- Web-based terminal that directly interacts with ContainerSSH via websockets.

- File manager using the same.

- Better audit logging for SFTP interactions, allowing for filename-based search.

- Web interface for viewing audit logs.

There are a few talks/videos available online:

- ContainerSSH in one minute [2]

- Building a science lab with ContainerSSH (CERN usecase, Red Hat Research Day EU 2022) [3]

- Build your own honeypot with ContainerSSH (DevConf CZ 2021) [4]

[1]: https://containerssh.io

[2]: https://www.youtube.com/watch?v=Cs9OrnPi2IM

[3]: https://www.youtube.com/watch?v=g7M7japaa1o

[4]: https://www.youtube.com/watch?v=aqJpsx7HpkU

I tried ContainerSSH and it is great. What are you doing now that you switched industries?:)

Thanks, glad you like it. :) My wife and I left the company with the red headgear and started a games company called Something Pink[1]. It's still a bit in the making, but we are uploading discussions on our progress to YouTube. The only use I have for containers is running an SVN server. :)

[1] https://something.pink

Here’s another project I worked on that I know will probably ruffle some feathers here. But to preface, I used it as way to learn Open AI function calls and - small concept of a zero to product launch full automation. This is a single page script that that generates an idea, creates an audience profile, then writes a book of short stories in a 5d array, looping over each level, then generates an image through midjourney through a hacked together api and creates a styled cover where there are a number of steps to ensure that aesthetically, the book cover is decent. It then uploads it to Amazon using puppeteer and uploads it to the woocommerce site as a new product.

https://github.com/mattrichmo/wisewriterv3

It’s my first complex app I’ve created that required multiple apis, and dealing with errors from multiple levels but I never actually uploaded a single book to Amazon so I can’t speak to the quality of the content. I was thinking if throwing this up on a website and letting people make a book for their child or something but I’d love the take the idea of zero-to-launch automation further. I’d love to go to bed and have a new well researched, highly desired product that actually serves a market need when I wake up sort of deal.

I'm working on a Chrome extension that adds realtime presence (cursors, chat, gifs, sticky notes) to any website or page.

Each website/page (depending on your preferences) becomes an ephemeral channel where you can see and interact with others users browsing the internet. It's still in a WIP shape but very functional.

Website: https://webcursors.click/ Extension: https://chrome.google.com/webstore/detail/web-cursors/chcedo...

This is a pretty cool idea to add a social element to everything. I don't think I'm necessarily in the target demo but still this looks really cool

Thanks, this is definitely one of the most enjoyable side project I've done so far. It's rapidly satisfying, I can easily add new features, and I find it fun to see and interact even with a few users on random websites. I've been sharing it a little around me and so far I've had some nice feedback, so I'm pushing it!

For about a year I've been going through the process of gathering all of the television listing pages from the New York Times, processing them to extract the films that played each day, matched the films to their IMDB links, and parsing the VCR codes to get the time/channel as well. Now I'm creating a little frontend to view and analyze the data as well as curate bits that didn't match properly.

At the moment I'm mostly just planning on sharing it with my brother, but who knows. I generally think I'll eventually share the data somewhere possibly on a website. I can't commercialise it if I understand the New York Times license properly since I scraped the Timesmachine for the listing pages.

Regardless, it's been super fun, and the data is very cool. Like 21 thousand unique films. At the moment I'm mostly trying to decide if I want to continue adding bits, like seeing all the listings a specific actor was in ... Mostly to test the hypothesis purported by PCU that there is always a Michael Caine or Gene Hackman film on television (I'm joking obviously there isn't although it is closer than you'd think). At the moment you can view an individual date or all of the listings for a specific film.

To me this is starting to feel almost like a Wikipedia thing or internet archive. This is, in a way, a specific ground truth of when television programs played in the New York area. I just have no idea who or what would be interested in it.

The place to put the program metadata is Wikidata, and to share and curate the schedules you could start a new wikibase e.g. here: https://www.wikibase.cloud/

Currently I am developing a browser extension to improve the review process on GitLab.

At the company, where I work, we are heavily relying on multiple reviewers to promote familiarity with the code base, as well as, of course, to increase overall code quality.

Gitlab Premium and Ultimate allow to use multiple reviewers, however, we don't need any other features from those tiers (well, maybe organization-wide code search could be nice as well).

As the result, I created a browser extension, which allows to use multiple reviewers via comments and descriptions. It also has graceful degradation, i.e. even team members without the extension will still be able to participate in the process.

On top of that the extension adds merge request deadlines with notifications, comment labels (viewable without the extension as well), custom review statuses ("Approved conditionally " is a real time saver) and more.

The extension works pretty well in our company, however, I would love to see if it's applicable to other companies as well, or what other features people need.

https://chrome.google.com/webstore/detail/codecode-review-fo...

https://addons.mozilla.org/en-US/firefox/addon/codecode-gitl...

https://mapofthebest.com

I love food and I love travel. Naturally I decided to compile a list of all the best restaurants in the world. Still very much a work-in-progress but would love to get feedback.

This is amazing. So much easier than the official Michelin app. I’ve had the same idea to aggregate real reviews… because google reviews are pretty useless if you’re looking for special places.

Bookmarked and will use!

Thanks! I struggled with the same problems. Existing search tools make it difficult to discover high-quality restaurants (either via poor UI, lack of granular filters, or low quality content). I'd be curious to hear if you have a better solution to filter/gather reviews.

Opened the site to see if I can provide some feedback. Ended up bookmarking 6 places to go check out in the city.

So glad to hear that! After building the V1, I was similarly surprised by how many new restaurants I was able to instantly discover. Even found a few spots which I'm tempted to travel for.

Like others, bookmarked. So useful, thank you. Feedback as requested:

1. It's a map of "the best", so maybe the default filter shouldn't show "ok" recommendations

2. The numbers on the pins change regularly (I'm on mobile/Firefox). Maybe the pin could indicate the source of the review, or the quality, or the restaurant type? Or not have the number at all.

3. Moving the map around reloads the pins for the viewed area, but it sometimes loads a different set of pins for a big city. Is there a max number of pins being loaded on each map view change? If so, perhaps the max should be higher?

Hope that helps!

1. Agreed, I'll likely hide them by default. The results from Gayot/Eater are incredibly noisy, and I doubt most people are interested in seeing restaurants from Gayot's "Top 10 dog-friendly date brunch spots this holiday."

2/3. I was experimenting with numeric pins but have now switched back to pins which vary by size depending on the restaurant awards. I also increased the pin count to 50 by default. Can you try it again and see if it feels more intuitive?

Awesome!

The pin count increase has improved the situation, but it's still noticeable. I'm using Vancouver/CA as the test if that helps.

I see what you mean now. This has become an interesting experiment in UX. I updated the platform to leave any previous pins on the map (even if they're no longer in the top 50 results). Hopefully it works better now!

Great work, TYVM. I'd suggest submitting it to subreddits like InternetisBeautiful, MapPorn, etc..

If you want to monetize it, setup ads and a "adblock" detection script.

Wow also had a similar idea! How are you getting the google review aggregate if you don't mind me asking? Even simply plotting Michelin restaurants seems like such a good idea in retrospect. I wanted to do something similar with famous food bloggers, maybe you can consider adding that to your list. Something like all of Mark Wein's recs on a map, with a link back to the original youtube video. Automatically adding data from different sources is the hard part I couldn't get over, especially since some of the videos don't have a location in the text.

Reviews are coming from Google directly. I'll likely add OpenTable reviews as well since I feel their content quality is quite high.

I'd love to add bloggers but it seems quite difficult to automate and I want to minimize how much manual work this project requires.

Similarly I'd love to include all the chefs and track which restaurant they're at now but I haven't found a good way to automate this.

Very cool - I assume most of the data is gathered through scraping? Quite a few of the sources you're pulling from don't look like they have APIs available.

I'm nearing the 0.1 release of Burro, a new typesetting language intended specifically for text-heavy documents. (If you've used the mom macro set for groff, it will seem very familiar to you. I like to describe Burro as a spiritual successor to mom.) The goal is to eventually have all the power of TeX with a much more pleasant syntax (after all, a typesetting language should be easy to type!).

The goal of the 0.1 release is to have a strong engine for setting English text, and I'm pretty much there. The main focus of 0.2 will be the macro system and beginning work on internationalization.

I started Burro in October 2018 and took a four year break (yay side projects!). I came back to it early this year and have been working on it pretty consistently ever since. It's exciting to have gotten this far!

Website: https://burro.sh

Source: https://git.sr.ht/~reesmichael1/burro

I wanted to make sure you're aware of https://github.com/typst/typst -- Typst is a great clean formatting language that's worthy of taking on TeX!

Typst might be powerful enough that Burro's "explicit layout" desire could be implemented as a library in it.

I am! To be honest, I was a little bit sad when typst first came out, because I'd already had Burro in the works for some time before then. However, typst and Burro are very different, so there's room for both to exist!

I love how you showed an input and output. I posted a language that I'm working on in this thread, and I'm definitely going to include that on my page.

I have been working on a web app that makes tutorials for elderly people who struggle with technology. Something as simple as "turn off the wifi and turn on data" can be a struggle for a lot of people.

My plan is to make tools for making tutorials customized exactly to the users phone / desktop screen. So if an icon is on bottom right, the tutorial should mirror the same.

I have sporadically worked on it for the few months. The only issue is whether I am dumbing down the app way too much.

My experience of building accessible systems for low-literacy user groups is there's pretty much no such thing as dumbing down too much.

But the one big lesson is user test, user test and user test again. People will get tripped up by all sorts of stuff you can't see coming.

The good news is it sounds like your app could be mocked-up cheaply to get it infront of people to test your hypothesis before you've spent too much time building stuff.

There are umpteen agencies around that can help with research like this, including finding appropriate cohorts of people and even doing the user testing on your behalf. It sounds like it would be well worth the money.

Thanks for the feedback.

> .. is there's pretty much no such thing as dumbing down too much.

Yes, you're right. I think it's easy to forget what's basic to someone, is often very challenging to someone else.

Neat. Much needed.

Yes and: I would've loved a "parental controls" for eldercare. Config and arrange the entire UX. Mostly to remove stuff, highly constrain the UX. I'd be fine with a shell. Like the olden days, when we used AUTOEXEC.BAT to launch AutoMenu or Norton Commander on startup.

While the constainly evolving UI/UX of iOS is probably a good thing, each new iteration completely defeats my elders. They're simply past the point of learning new things.

For a device that champions a non-modal paradigm, the iOS UI maze has a lot dead-ends.

They should also do usability testing of Siri with elders. Something as simple as a "start over" would be great.

Is it too developed if it’s in private alpha?

Anyway … I’ve been working on CodeApprove (https://codeapprove.com) for a while because I think code review is the biggest shortcoming of GitHub.

The idea is to make a code review tool for power users. Just like you have an editor that makes you productive at writing and reading code, you should have a review app that makes you and your team productive at reviewing it.

Or for any Xooglers out there: imagine critique but on GitHub.

Would love feedback on how it’s positioned and if I could improve the value prop!

We use BitBucket at work, so I won't try it out in real-life conditions, but I agree that there seems to be a niche there. The default code review experience is very poor!

This is really, really cool. Great landing page—solid design, effectively sells me on the features and benefits.

Would love to see it in action—either some video or a way to try it on a private repo. Even if I can’t actually submit a PR review. :)

Thanks for the feedback! I definitely need to get over hating the sound of my own voice and record a demo video.

You could try to use an AI text-to-speech to get a different voice? I wouldn’t judge based on voice (or the lack of it), but I also understand hating the sound of your own voice when recorded. ;)

Maybe even just short gifs/videos that show features in action alongside where you currently have some images?

Looks cool. With it being in "private alpha", if I drop it in a Slack for an open source project with which I am involved, and the maintainer/leader wants to try it, will he be able to sign up?

Good question, and yes! I'll get a text message as soon as they ask to sign up and I'll have them in the alpha by the end of the day. Anyone who is already in the alpha can invite unlimited teammates.

It's probably ready for public beta ...

I agree that a better tool is needed! I like all the improvements you are selling the tool on, but one of my biggest gripes with code review experiences has been the inability to specify the ordering of the files in the diff. Most people look at the diffs from top to bottom by default, and I sometimes have my questions answered when I move down in a diff. It would be so much better for the author to be able to select an order.

Good luck!

Oh do you mean allowing the author to select the order? That's a very cool idea!

I've wanted exactly this before - would be great to easily tell a story in the code review. Or an easier way to just see each commit in isolation (with a previous/next button).

Have been working on a project which allows you to check if any of your source code has been copied from major online projects (over 70 billion lines of code) and report back if duplicates have been found. You can control the length of the run (so 15 lines means a match or 7 lines) and make the line matching fuzzy.

The idea being to provide compliance checks such as licensing and such, although I guess it would allow for checking for potential source code leaks as well and as a plagiarism detector.

It works by running a program over your source code which produces line by line hashes, with the result uploaded and then scanned over. The result is a report indicating what percentage of code was identified as copied both on a project and on an individual file level, with the identified source and its location. As such if nothing is leaked there is no ability to reconstruct the source code as it is a hash.

It works over 200+ languages, and while working over line by line currently can be expanded out to work in different ways.

Looking for interested parties to try it out (see my bio for contact) in addition to the current ones I have. Even if its just you wanting to kick the tires.

I think there is a company that does this exact thing.

If you happen to know the name I’d love to hear it. I suspect I’m not searching for the right terms.

I'm working on https://commitspark.com , a headless CMS where all data is stored in a GitHub repository with branching/merging workflow support for content.

Reading/writing content from/to GitHub happens through an Open Source API library I released. For the content data schema, I simply require a plaintext GraphQL type file inside the repository. The schema then automatically determines the API structure as well as the editor-friendly UI that I can generate on the Commitspark website when you log in via GitHub.

Even though Commitspark is already publicly available, it is still 100% a side-project that doesn't earn me any money (yet).

Vulgar computation.

The vulgus in Latin were the common folks, the non elite, hence referring to cussing ass “vulgarity,” as in “we nobles do not talk that way, that is how the common folk talk!”. And if I do my job right most professional programmers should be offended by the result? But that's a really awful thing to have as your north star.

The basic idea is that computation should be something that does not require a degree and work experience and other markers of professionalism... Computation is just an intellectual amplifier. I was shocked some 5 to 10 years ago to have gone to a guy’s “Excel Tips” meeting at the advertising firm I worked for—all the other programmers skipped—and saw a bunch of non-technical folks excitedly sitting down to learn what I rapidly realized was... programming! They were volunteering to run and sit at a programming class!!

But Excel is a shitty language, but it is also vulgar. When you go to learn it, there is no compiler step, the interpreter step is a repl of sorts, everything is being automatically printf-debugged for you by default, and helpful colorful boxes point to the data dependencies of the cell you're working on while you're working on it.

But it's also like, why isn't everything like that. If I give you a UI that is written this way, you should expect to be able to long press on a button and see the button disassemble before your very eyes, you can track the pipeline for its clicks, and if that button doesn't do what you want it to do, you just change what it does. Everybody on this planet should be able to do that, it shouldn't require ivy league ivory tower education bullshit.

So I have these prototype languages that are fundamentally graphical, and kind of amount to a sort of XSLT (of all things!) being compiled to WASM?. A bit of inspiration from hypercard here and there. And it's weird because the decision to WASM is kind of its own shit decision? Like, I wanted the ability to just text you an app that I am working on and you can run it on your phone and feel safe that I am not stealing all your secrets, was why I started compiling to WASM. But as far as I can tell there is nothing I can do with WASM other than to define a function library, so the resulting language such as it is, is mostly a bunch of conventions operating on a WASM function library, and it gets really elaborate and complicated and I am 90% sure that you could just exploit the conventions in the state that it is right now, so like “why am I writing XSLT-to-WASM who the fuck cares” sometimes.

Nothing fully worth showing off right now though.

Is it a thing like htmx and _hyperscript?

I love the concept (and the name too) of vulgar computation!

This is a really interesting idea. If software was repairable/hackable like hardware...

I’m working on developing a new kind of flexible linear actuator for robotics and prosthetics. Essentially an artificial muscle. If we can make production economical it’ll make a big difference in how energy efficient and human-like robots and prosthetics can be. I think it’s plainly evident that traditional electric motors and servo actuators are a big limitation on robots in how well they can mimic human level functions, like grasping and walking. A robot might only have a handful of actuators in their legs to walk, if evolution could have found a way to walk proficiently with so few muscles as their are actuators in a robot like the Tesla bot, it would have. Can’t speak too much about the specifics of how it works, but its using electromagnets, no pneumatics or electrostatics or shape changing polymers etc. I have interest from a few uni research groups in my city, but we’re always on the lookout for groups who might be good alpha or beta candidates. My email is zak at zerocubic dot com

https://github.com/mickael-kerjean/filestash

This is what I wish Dropbox was, a simple layer that make interacting with your FTP server easy so nobody has to own your data. The end game is both to be feature complete with Dropbox and be able to change every aspect of the application through plugin so everyone can get out what they want from it. To do that, I’m making the core a sort of framework to build file manager like application.

https://imgur.com/a/i0pgsRx

What about TBs transfers over Internet? People like to copy VMs..

Does it have any form of brute-force/DOS protection or a way to slap fail2ban for it? One of the reason I would replace our IIS FTP with this, it's constantly under attack and I'm too lazy to write another bruteforce protection script.

I would put this behind a VPN to keep it private. I have used this[0] Wireguard setup before, which takes the grunt work out of configuring everything.

[0] https://github.com/notthebee/ansible-easy-vpn

Thanks for the suggestion, but this needs to be public, otherwise it wouldn't needed a bruteforce protection.

* Tool that creates Twitter (X?) / Instagram feeds to RSS so I don't have to deal with algorithms etc.

* A turn based game that is essentially "football manager for counter strike" where you control the whole team and their actions in "planning" and "execution" phases.

* I run a Slack community for the South African tech scene, constantly hovering between "I wish I could sell this and move on" and "I really should put in the effort to monetise / grow this".

For point 2, ever seen Door Kickers? https://store.steampowered.com/app/248610/Door_Kickers/

This is great, thanks for the link!

I was using 1, until Twitter shut down their API. How do you do this now? Scrape the HTML?

Yeah or use "unofficial apis" e.g. Nitter (https://github.com/zedeus/nitter)

Unfortunately Nitter broke earlier this week thanks to some changes by Twitter: https://github.com/zedeus/nitter/issues/983

I was using a self-hosted Nitter instance for the RSS feeds as I deactivated my Twitter account

How do early stage Tech enterprises deal with the instability and problematic infrastructure in South Africa?

This is quite hard to answer in any succinct way. It mostly comes down to "you're earning enough to just get around them".

SA is kind of a place of extremes, there is extreme poverty, unemployment etc. But if you're lucky enough to have a good job you get Uber Eats, Netflix, 200mb fibre to your house etc.

I built mirrored fables, an AI powered visual novel generator. Give it a try!

https://www.mirroredfables.com/?proxyKey=hellohackernews

source: https://github.com/mirroredfables/mirroredfables

I love the idea, but I keep on getting hung up (after 10+ minutes) on

``` [worker] Searching for music... ```

If it helps, my prompt was:

> I would like a story set 2000 years in the future, where humanity has achieved interstellar travel. Humans prioritize exploration of extra-solar planets and the search for intelligent life

Oops, it seems that I had hit a youtube API rate limit last night. Should be okay now.

I enjoyed my story!

It was based on the prompt I shared previously.

The protagonist was Aeyela who was rendered as an anime/manga school girl. Her voice was portrayed as deep and masculine, which was pretty entertaining.

I did encounter a couple issues:

* the 2nd song was age-restricted, and I was unable to skip it, so the music stopped. I looked at another tab and returned, and another song loaded

* potentially related to the above: the voice narration stopped and I couldn't figure out how to get it to resume, even after toggling a few of the related settings

Love to try but I've been stuck on [worker] Searching for music...

Oops, it seems that I had hit a youtube API rate limit last night. Should be okay now. :)

I love this, but looks like it's less VN and more of just a novel.

I think the problem is it generates content, but not very high quality content. It's not gripping enough from the beginning, just sort of drags out. It could be that the engine works right, just that the input is garbage, as often happens with AI.

But maybe something about it could be engineered a little better. From the code it looks like it's just generating most of it in a single blast. But I wonder if it can be iterated into chunks.

e.g. split it into the Three Act Structure.

1. Protagonist really wants something. Faces an obstacle. Tries everything and gets stuck. 2. Protagonist makes sacrifices. Gets even more stuck. Something that the protagonist faces in Act 1 ends up being the key, often a mentor or a sacrifice that Protagonist didn't expect. 3. Denouement.

You could probably generate a want and an obstacle. I kind of did a prototype of that here (feel free to steal ideas): https://random-character-generator.com/

Then do the same with the other characters. Generate relationships.

Then some iteration where the AI writes those characters into those scripts.

Probably another where it writes up scenes/places/setting later. The characters are then colored in and put into these places.

An editing run where it removes unnecessary scenes or scenes that didn't work. Or leave this to the humans.

Anyway, it is very good work. It's nice that it's open source, may contribute to it some day.

Haha thank you for the encouragement!

You have got a nice generator there! I am very curious what's powering it, as it seems to be instant. I dropped you an email, let's chat!

Re: structure and cohesion- this is where GPT4 vastly outperforms GPT3. For example, it is much better at using a hero's journey and managing plot twists. You can go into the AI settings window and set the model to use GPT4 instead [I'll leave how to get there as an exercise ;)], and give it a try.

Currently the generator works in a very similar way to how you described it: 1. It takes your input and generate a world setting and the characters in it, including relationships 2. Based on the information from above, it generates a overall plot outline, broken down into chapters 3. Based on the information from above, it then fills out the chapters with dialog

Currently the biggest limitation is context length and cost, and I haven't focused on optimizing that yet.

You are absolutely right that a human editor would vastly improve the quality. You can see a lightly guided early pre-generated example here [https://www.mirroredfables.com/?gameUuid=302fe7bf-4ff0-4a77-...]

An AI story editor is a very interesting idea that I haven't thought of before, and it makes a lot of sense. I am going to play around with it!

I started working on a musical web app (https://synth-app.vercel.app/) using Tone.js that allows people to easily create grooves and melodies. It is functional, but I'm having some issues with the beat being irregular with certain numbers of beats (not sure if it is purely the AMOUNT of synthesizer objects I'm creating or if it is just a problem with odd beat numbers).

I also have plans to add a feature to save and share your projects, but not implemented yet and I'm always curious what other features or musical capabilities people would enjoy playing with.

I'm also working on cli time tracking tool that is still very basic but I'm sort of wondering how best to add more functionality. The current main branch allows you to start a task with `$ start -t` and you can add details and then list all the events you've tracked, all saved to sqlite database. I like how immediate it is to track things from the command line, but I'm still thinking about how to display the events and tasks you've tracked (I'm building with rust and started tinkering with Ratatui to build a TUI, though I prefer just having a cli most of the time). I'm also still thinking about cli UX. I kind of prefer the git hub cli style that has commands and sub-commands because it seems more orderly than just having long lists of arguments. Anyway it's very in the beginning stages but would love recommendations on reading for UX for command line interfaces. https://github.com/farmeroy/Time-Bandit

Looks nice! I made one of these, but with drum machine samples instead, back when <audio> elements were the new shiny thing.

You might want to look into adding a DynamicsCompressorNode near the end of your web audio stack; it should help with the clipping issues you're getting whenever there's polyphony going on.

Thanks! Yes, the sine waves really build on each other and have unpredictable volumes. I'll try the DynamicsCompressorNode out

https://github.com/lukesalamone/code_nitro

Convert your code to an image from the command line. No internet connection needed, although it can also download and convert code from Github too. Made it because currently the best alternative (carbon-now-cli) sends your code over the wire and there's a lot of situations where that's not desirable.

This is quite nice! I might start using this.

Some things that would make this nicer:

- Add a downloadable, ready-to-use bin as a release. - Change the background-gradient. For me, it is a bit too close to the bright pink used on old VGA-games, so I initially thought it was an error. - Add option to also copy those lines to clipboard, or upload them to pastebin or similar. The url could then be embedded within the generated image. Perhaps this does not fit into the scope, but it could be nice. - Add a short instruction on how to use this within an editor. For instance, perhaps a short vim-binding from a selection could be added to the readme? - Why is there a link in the readme to a wikipedia-article on Fast inverse square root?

Thanks for the feedback! Yes, the installation process is a bit barebones at the moment, a pip install would be a lot cleaner.

As far as the wiki link, it's a reference for the code sample I use in the demo.

https://apiforllm.com/

I've been working on this on and off for a few weeks now. The idea is to let LLMs interact with the outside world using function calling. Eg- I built a simple version of chatgpt web browsing using function calling. There's a django site which is responsible for the auth, managing chats etc and then there's a flask site that's responsible for running functions in a gvisor sandbox. Users will be able to upload their functions as docker images and I'll run them in a sandbox.

There is a lot of self doubt at this point: who is this for? Will people pay for usage? Should I open source the entire thing or part of it? To add to this: I have no online audience, how do I promote this?

I do have some answers to these questions. Eg- should I open source the flask site and get llama 2 to use it? It provides value to the community + hopefully I can get some attention for the project.

I would appreciate any feedback on this project. Please send me an email if you want some openai credits added to your account.

Edit: use testaccount/demouser to login, I've added some credits for testing.

Hey, so I got pretty far into implementing open api specs for LLMs. To the point where you can point to an open api spec, ingest it, and then it’s available for GPT to select to satisfy user requests.

I do this by minify the specs and building an index to feed it that it selects from using logit bias.

Im at pause point on it while I figure out how to actually have GPT build the calls, but I think something like you have might solve that.

https://github.com/ShelbyJenkins/LLM-OpenAPI-minifier

Before the function calling update[0] it was possible for the gpt models to use tools using specific prompts but it was unreliable. Have you tried using function calling to let the model build the calls?

[0] https://openai.com/blog/function-calling-and-other-api-updat...

I am working on an experiment called UltraStrong, completing an Open Strongman Competition, and running a 50 miler Ultramarathon over a single weekend. I have done each separately but can you do both simultaneously? During the process I am keeping a journal, documenting everything, with the idea of writing a book at the end in a similar style to Henry David Thoreau's Walden.

What an endeavor! As a 2 hr half marathoner and some with a 100 lb bench, I am in awe.

I have been working on a time killer mobile-first geometric game with a novel gameplay.

There are vertical light rays, and optics transforming them are falling from the sky. The goal is to get rid of all the darkness.

Preview:

https://twitter.com/klntsky/status/1687853518395240448

APK: https://github.com/klntsky/SunKey/releases

Looks trippy, but the animations/colors need a bit of work. The Twitter video made my eyes hurt.

How in particular do you think colors should be improved?

I think the game needs more contrast. Make the blocks that change the stream a different color and not transparent maybe?

They need to be at least partially transparent to show how the geometry changes within a prism.

https://www.pgrammer.com/ Coding interview practice questions like LeetCode but using GPT to analyze the code and provide hints when you get stuck. It will also look over your submission and give feedback like an interviewer.

I've conducted A LOT of interviews and have been the interviewee plenty of times. To me the hardest part about prepping on LeetCode was that I would sometimes get stuck on something. Not necessarily stuck to find any solution but the optimal one. The hints I'm getting back from GPT have proven to be very valuable and it still lets me code the solution with those pointers.

It's not a perfect product by any means but it's my work in progress passion project that I believe has a lot of potential.

Super fun to use. I hate having to “give up” on a question when practicing and this gives me a bit more options rather than just pass/fail

yes! Part of prepping is to gain confidence and it can be very discouraging to run into this before live sessions.

Oh nice, this could be my new favorite thing. It's less lonely than the usual. It's great for warming up, basically just trying to code from memory instead of relying on docs.

I love that it yells at me for naming my variables "xxx" too. I also love that it goes all "Candidate did not follow instructions" when it tells me not to use a specific function, but also tells me that the solution would get the right answer. It's just one of those things you wouldn't try on a real interviewer, but want to try to see if it works.

But $96 per year is quite a commitment for something that I could get bored of next week. Would be nice if there was a monthly option.

Thanks for that feedback! I will add a monthly option today.

Hey, IDK if you'll see this, but is there an email where we can send more extensive feedback or generally contact?

Hey! Yeah my email is in my profile. Feedback is super appreciated!

Let us know when there's a monthly option. Just waiting for that to subscribe.

Will definitely try this. Right now, I am finding ChatGPT a great tutor during coding interview prep (I don't really use it for hints that much, personally; I've mostly been asking it to explain the optimal solutions to various problems, because the Leetcode editorials are usually pretty poorly written, and in any event I want to be able to ask questions about them).

One thing I would like to add as a pain-point when studying for algo interviews is that I want sources of questions other than LeetCode. Maybe this has some.

I've been working on blog powered entirely by GitHub Gists. https://gist.aviperl.me/

With zero extra effort, you get a simple blog. With configuration via settings files stored in a secret gist or on a gist itself, a rich blogging experience.

A currently hidden feature that is useful is RSS feeds for any GitHub users gists: https://gist.aviperl.me/avi-perl/rss

I'm hoping that the GitHub graphQL API might let me get list of gists with their file content, in which case, no outside build step would be needed to provide the full experience for a GitHub user.

What I mean here by "full experience" is adding a title, image, tags, etc.

This is beautiful. I might start using this!

If you provide me with your username and the id of a secret gist for configuration, I'll add your username to the list of users to precompile pages and use the full features.

This is a neat idea! I had a similar idea but with github repos instead of gists - it'd be possible to use tags to just check out blogs you're interested in. I built mine as a neovim plugin rather than a standalone site though. Gists definitely make it easier for shorter blogs.

Hi everyone :) I'm a student at UWaterloo in Canada. My friend and I made a site that can automatically dub videos to different languages: https://www.dubbah.co . We use Whisper with Speaker Diarization, GPT and some voice cloning models. Would love to answer questions :)

One thing to try is speaker identification in dubbing (different color subtitles)

Hi! Someday, I feared Pocket [1] would disapear and I would suddenly lose all links I saved there. So I made pouch [2]. It let's you save links and, if possible, automatically saves the text, for a reading mode.

I never showed it publicaly, so I'm the only user. Fell free to use it and I'm open to any feedback.

Mind you I'm a backend dev, so in terms of design and overall frontend quality it is definitely lacking.

[1] https://getpocket.com/pt/

[2] https://pouch.lucashmsilva.com/

https://github.com/achembarpu/pockyt

"When will I run out of money?" https://whenwillirunoutofmoney.com/

Due to changing jobs recently I created this site as a fun side project, as other sites I found involved sending your financial data to someone else's server. But I'm not really happy with the introductory text, nor did any friend I showed it to find the "help" link which was disappointing. Would love some feedback or advice!

I love the idea of a 'sheet' where you can add financial data and have it figure out what it needs to calculate.

I would love to see this have the ability to draw simple graphs to show how your situation will look, or allow for things like interest rates on deposits, or the ability to see how your income might look like in X years with an Y% average increase, etc.

PS: Also check: https://soulver.app/ for inspiration

> I would love to see this have the ability to draw simple graphs to show how your situation will look

That's a great idea. One tool I had been previously using (server based, then started to require payment so I stopped using it) had a graph. I found that while it was cool initially in the end I just looked at "do I have enough money to make it to date X", so I figured I'd skip implementing a graph. But maybe I should consider it, it certainly helps to visualise the situation.

> or allow for things like interest rates on deposits, or the ability to see how your income might look like in X years with an Y% average increase, etc.

Thanks for the suggestion, I thought of that too, wouldn't be hard to do...

This is amazing. Didn't even need any documentation. It works with how I track money, mostly in evernote and not spreadsheets.

I wish it did other periods like weekly, daily, annually. Weekly is useful for tracking food. Annually is useful to decide whether to cancel those subscriptions, or converting them to monthly.

Might use this for taxes too and change the formatting of how I track money to match this.

Thanks for the ideas, and also thanks for the explanation (it's tempting to think nobody would need feature X, but when you put it like that I can totally see it makes sense!)

There is "yearly" already but I think your name of "annually" would be much better, so I'll add that naming, thanks!

Going to be piloting in hyper-local form first (one deanery in one diocese), but I'm working on a sort of Nextdoor-meets-Public-Square for Catholics.

First just a list of all local Catholic events with subscribe-able tags for target audiences, but then also a directory of local businesses run by parishioners.

Maaaaaybe eventually message-board-type features, but that's far enough in the future to ignore for now.

No idea yet if anyone will care about it as much as I do, but it's something I wish existed, and nobody else seems to be building it!

1) can it recognize handwriting? 2) can it identify where a child made a mistake when ‘showing their work’ aka error analysis?

A tool that guides a kid to their mistakes would probably be more useful than on that solves a problem on their behalf.

It seems you were meant to reply to francescogior instead.

You are quite right.

Good, 'cause I was really confused there for a second!

https://git.sr.ht/~rockorager/vaxis

A(nother) golang TUI library. Targeting modern terminal features, and a few different APIs for building applications. I really like notcurses, but the CGO cost when setting individual cells is very high...so this was my attempt to "port" what notcurses does to go. I handle things a little differently than notcurses, but in general am taking a lot of the same approaches for finding out terminal capabilities. Just not nearly as many optimizations when it comes to the render side (but Vaxis can render at about 5000 FPS in typical cases, so I haven't tried to optimize anything more there than it already is).

The main design decision I am going back and forth on how much I have used globals. Initially the design was a big struct, but I liked being able to use globals so when I write widgets as part of the library, they can send messages into the event queue without needing access to an application struct or something. I can't think of a single use case where a user would need more than one instance, so globals have seemed fine (and even notcurses says you can have only one notcurses instance per process...so I'm not even straying from that).

Thanks for any feedback!

I'm a Data Scientist. For some time, I've been working on a library for feature engineering. • GitHub: https://github.com/feature-express/feature-express • Website: https://feature.express It isn't yet complete, and I wouldn't consider it ready for production use or handling larger datasets. Here are some of its characteristics: • Event-based workflows: Initially, everything is converted to an event format, ingested into an event store, and processed from there. • In-memory: Both the event store and evaluation have been built from scratch. • Written in Rust, but there's a Python package available. • A DSL (Domain Specific Language) for defining aggregations, similar to SQL. Why am I developing this? I've always found it challenging to build models based on time. These models can be surprisingly tricky, and there's a high risk of accidentally using future data, which can lead to data leakage. FeatureExpress is designed to nearly eliminate such mistakes. Moreover, I believe that representing data as events is an intuitive approach.

A series of travel-based puzzles to solve as a way to explore a new city. Going to Edinburgh? You get a series of clues to solve, using landmarks in that city. It will take you to many of the areas a tourist would want to visit, but in a more engaging way than a list of things you 'must see.'

I'm trying to nail down the exact audience/tone. Is it for families with kids? How hard should the clues be? Should it be more lighthearted or immersive?

I am making https://gpxgo.com, a route planner for road cyclists.

I am aware of the existence of strava and komoot but I had some frustrations creating specific routes using their apps on mobile. Especially adding and dragging waypoints to a route was a focus point. In addition to exporting the route to a gpx file, the website allows to view the elevation profile and road type.

This is really interesting as I am looking for a route planner that I can use with my old Garmin Edge without having to do 3rd degree API connections in the background every time.

All I want is a route file that I can copy on the device using a file manager.

Can this tool work for me or do you know other bike computers you were able to use this with?

If your old Garmin is compatible with the Garmin app you can download the gpx file then open it with Garmin app and sync it to your Garmin edge. Not sure if moving the gpx file to your Garmin using a file manager will work.

It will work, you just need to put it in the right place on the Garmin (“NewFiles” or something like that).

This looks cool, but please add translations, I don't speak Dutch(?). :) One question: are you using OpenCycleMap? If not, it would be worth adding as it highlights a lot of the safer routes, such as Eurovelo and local bikeways.

Yes it's Dutch, will add translations soon. I like the OpenCycleMap suggestion. I am indeed looking for ways to highlight safe or nice roads by using OCM or maybe Strava segment data.

I've developed a niche app designed to assist individuals in the Dominican Republic and potentially others outside the US who purchase from platforms like Amazon US and eBay US in retrieving sales tax refunds. The way it operates is by accessing data from your Amazon purchase history and your courier or forwarding service. From this, it auto-generates an email to Amazon to reclaim the sales tax. Initially, I aimed to introduce a commission-based model from these refunds, but this aspect is still in the pipeline.

The tool did function for a short while, during which it successfuly reclaimed a few hundred dollars in refunds. However, it broke when a courier site introduced reCAPTCHA at their login and Amazon made UI changes to their order page. I'm now at a crossroads, contemplating whether the upkeep required to ensure the app keeps working is worth my while, given how niche the potential market is. I am not sure if I can make realistically make money from this. Would love to hear insights or suggestions.

For reference, you can check it out at https://taxrefunder.io, although the core functionality is down at the moment.

You absolutely can. I recommend selling it as subscription service at launch, with a free demo. Less calculations/simpler workflows. If it allows to save $hundreds, people will be happy to pay $tens. A lot of "lifehack" websites accept paid content, you even have a chance of getting free advertising.

Why isn't your product global from the start?

> A lot of "lifehack" websites accept paid content, you even have a chance of getting free advertising.

Thanks for the suggestion.

> Why isn't your product global from the start?

I'm not sure I understand the question but mostly because I have to add support for shipping forwarders (e.g. write scrapers for them) and right now, the two ones I am personally using only ship to the Dominican Republic.

I also don't know if people actually do this outside the Dominican Republic. It only solves a problem for people who ship from the US through a shipping forwarder. If you ask Amazon (and other e-commerces) to ship internationally, they usually don't charge a sales tax in the first place. I don't really know how common it is for people to ship through forwarders outside the DR to be honest. It's a big thing here because "last-mile delivery" is unreliable and it's easier to go at a shipping forwarder's local office to pick up packages.

http://polymorf.me/

A Talking Avatar AI generator similar to d-id. Worked on it for 2 months but I have problems adding natural head movement to the speaker. About 100 users so far but 0 revenue. Very low traffic (like 10-20 users/day). Not sure where to go next, at moment trying to focus more on marketing and getting feedback

Have you tried just going image-to-depth map? You could render the image in 3d with shallow depth then just shift the camera a bit

The main issue with moving mouths is the relative stillness, it's enough to introduce coherent motion in the user's field of vision.

If anything it might give you a unique distinction vs d-id, the head movements it doesn't aren't natural at all, but it's enough to not make it feel stiff

Interesting. I guess camera movement on 3d scenes would make the video more dynamic. That will put the focus more the "video aspect" as opposed to "Avatar" aspect of the product. I might explore that feature a little bit.

=> https://www.facebook.com/groups/virtualbeings

Neoq (https://github.com/acaloiaro/neoq) is a background job processor for Go.

It began from my desire to have a robust Postgres-backed job processor. What I quickly realized was that the interface in front of the queue was what was really important. This allowed me to add both in-memory and Redis (provided by asynq) backends behind the same interface. Which allows users to switch between different backends depending on their durability and performance constraints. E.g. in-memory for testing/development/low durability requirements, postgres when you're not running Google-scale jobs, and Redis for all the obvious use cases for a Redis-backed queue.

This allows me to swap out job queue backends without changing a line of job processor code.

I'm familiar with the theory that one shouldn't implement queues on Postgres, and to a large extent, I disagree with those theories. I'm confident you can point out a scenario in which one shouldn't, and I contend that those scenarios are the exception rather than the rule.

I am creating a documentation management software, with emphasis on a diagramming and system building - similar to Draw.io [1] but with better support on PDFs/Files, text/markdown/Latex, web links, and Search.

It is a desktop application and self-hosted, which will allow to sync with your work laptop and mobile devices without the need for 3rd party server. Many online solutions for knowledge management did not make me happy (e.g. Notion[2], Miro[3], Evernote[4]) as they all lacked either diagram building or were online only. There is also an Obsidian[5] for it is too complicated and it's unbelievable that a user has to go through hours of tutorials on youtube to understand how to use PKM software. Hence I created my easy but powerful tool for keeping all the knowledge/documentation in one place.

The desktop app was build using Kotlin Multiplatform (JVM app) with Compose UI framework

You can view(and comment!) the app screenshots here [6]: There are also mockups with more features, so ignore weird UI at first.

[1]: https://www.drawio.com [2]: https://www.notion.so [3]: https://miro.com [4]: https://evernote.com [5]: https://obsidian.md

[6: App screenshots]: https://drive.google.com/drive/folders/1lOIfGaXKbymw7fQXjwBh...

I understand this isn't the direction you're going in, but I would love a desktop app which could edit documentation (mainly text, but with diagrams), but save it locally to be checked in to Git. Imho documentation should be near the source, and go through the same branching and PR workflow that the source it describes goes through.

Ideally the desktop app would be a GUI which was easy to use, providing the basic features of Word or LyX or a markdown editor such as basic formatting, bullets, headings at various levels. Ideally some kind of diagram editor as well.

Ideally it'd save in a format which was reasonably amenable to being saved in Git and being merged from various branches, and which wouldn't be too verbose if you're looking through a PR on GitHub etc. So a text format, with no "# Created on <date>" fields. I have to say LyX solves most of these points so that's what I use at the moment.

> documentation should be near the source I totally agree with this statement. Moreover, I plan to have a feature that will allow to place your whole project folder into your vault in my app and then reference a diagram through the code/inner file (e.g. using comment blocks). And that file will be visible, searchable, within my documentation software. Think of it as a layer above the code.

However I do not want to tie the documentation process to git as it will be harder for other team members (think of POs, Teamleads, clients) to find the documentation through git history. Instead, a separate IDE for documentation (as is my app) is an ideal place to store & share knowledge. Just like there is separate app for slack, e-mail, browser, etc.

Thanks for your thoughts!

I developed a platform called https://rolematch.me, designed to assist job seekers in tailoring their resumes to match specific job advertisements more effectively.

The platform employs a custom-trained Language Model (LLM) to perform a semantic analysis of both the resume and the job description. This analysis then generates practical suggestions that could enhance the applicant's likelihood of bypassing Applicant Tracking Systems (ATS). Ultimately, the goal is to transform a job advertisement into a potential job offer.

Currently, rolematch.me is in its private beta phase. It does not collect or store any user data and is available free of charge. However, we are planning to introduce a premium version that will offer additional features such as cover letter assistance and job search capabilities.

For those interested in testing the platform, the invite code is the *name* of a well-known early cheat code. This code could be activated in the game Gradius by pressing buttons in a specific sequence.

Hey that sounds similar to what I’m working on![0] Neat to hear about other implementations.

[0] https://resgen.app

I got tired of tracking the status of my job applications in spreadsheets and emails, and built Rolepad (https://rolepad.com) to help with that. The goal is to help applicants focus on their job process and not expand mental energy on the mechanics of tracking dozens (and sometimes hundreds) of job applications. It's got a few beta users and I've gotten positive feedback, but there are still a few things I need to add before its ready for primetime (aka Show HN).

The tool is free and helps you manage your job applications, pipeline status, interviews, emails, notes, etc. Notable features include the ability to leave detailed reviews of your job process (+ read others' reviews and get various per-company stats... this is new, went live yesterday - tracking more fine-grained data than Glassdoor and others), forwarding emails to save@rolepad.com so they can be tracked as notes within Rolepad, sharing your progress via Sankey charts, and more.

What if your business model does not play out? Are you going to start charging fees?

https://chrome.google.com/webstore/detail/jpt-chatgpt-code-i...

I've been working on a chrome extension that let's you run python code in ChatGPT. You can see the output, see plots with matplotlib, upload and download files, use a bunch of packages, all thanks to Pyodide[0]. Basically ChatGPT's own code interpreter but worse (it's also free, you get what you pay for). It was important to me that it works completely locally and does not save any data.

It is one of my first personal projects and seeing the number of users go up makes me happy and want to keep improving it. I had to make my github repo private because I had an issue with a copycat, but I can make it public if people actually care.

[0] https://pyodide.org/en/stable/

This is very cool! I just downloaded the extension and tried a hello world program. Will tinker a bit more.

If you can make a library to run python code inside a JS project (like Next JS) then we could build a code interpreter in Next JS. Would be super cool. Here is a python implementation - https://github.com/shroominic/codeinterpreter-api - for your reference.

> If you can make a library to run python code inside a JS project

This is actually exactly what Pyodide does! It's basically the CPython interpreter compiled to wasm. This will allow you to run python code in, say, NextJS.

I'm trying to solve the problem of "how can I trust an e2ee messaging app on the web". Basically, the issue is that while e2ee messaging apps (think WhatsApp, Signal) assume no trust in the server, the user still has to trust the client -- and on the web, the server controls the client. Desktop and mobile platforms solve the trust issue in multiple ways: code signing, app stores, reproducible builds, publicly available hashes etc. On the web, none of that's possible. That's why Signal doesn't have a web client. WhatsApp does, but using it defeats the point of e2ee.

My proposed solution is to use Service Workers to cache a web app in the browser and employ clever tricks to prevent the server from pushing updates to either the Service Worker or the caches. This way the user can then control any updates and verify new versions using means that are already familiar from other ecosystems: comparing hashes, trusting only signed code, etc.

The goal isn't to develop a new e2ee messaging app. Instead I'm prototyping something that resembles an auto-update framework like Squirrel [1], only for web apps. Ideally it will be largely "plug and play", i.e. you could take the existing WhatsApp web app, serve it using the updater framework, and your users would now have a trustworthy version of WhatsApp.

So far I have a small amount of PoC-level code validating several small parts of the larger concept. For instance, I'm fairly confident that I will be able to reliably prevent forceful server-controlled updates, which is a core requirement. Right now I'm in the process of formalizing a threat model, hoping to spot any gaps before I move forward with the implementation.

Feedback on the idea in general would be highly appreciated, but I'd also love to hear any more specific concerns regarding technical solutions, UX, etc.

[1] https://github.com/Squirrel

I just started working on this last week at Recurse Center — a live coding playground for audio effects: https://fxplayground.pages.dev

The three main todos on my roadmap are:

- support languages other than JS (C/C++ would be the obvious ones)

- improve the visualization

- improve debugging (e.g. if you have a syntax error)

I had (have?) a side project to build a self-hostable Mint/PersonalCapital/whatever on top of ofxtools. The basic idea would be full control of your credentials, so you don't have to trust a third-party like Plaid or Yodlee. Self-hosted doesn't necessarily mean not supporting SaaS-ish behavior though, it's fairly easy to do "dump all my transactions to BigQuery in my own project".

I feel like self-hostable asset and expense tracking is something a few people would care about. I believe tech forward folks are faced with a choice between:

- Command-line tools like gnucash or ofxtools that are pretty bare bones

- Spreadsheets that are manually updated

- Give in, and hope that these companies don't have data breaches

Those options kinda suck :). Does anyone want this? I made it part of the way, until I got infuriated by the sad state of "open banking" in Europe (a friend was interested, but is in Europe, where OFX isn't popular).

A UI that includes what you’re describing could work on phones directly, self hosted locally on device.

I think more ppl than I imagine people have taken their run at this problem for personal use, including me.

It would be interesting to see if it could simply a few things I am parsing manually.

I’m working on ModelFusion, a TypeScript library for working with AI models (llm, image, etc.)

https://github.com/lgrammel/modelfusion

It is only getting limited traction so I’m wondering if I’m missing something fundamental with the approach that I’m taking.

This looks awesome! Keep sharing. I think many newbies are looking exactly for something like that including me to get startet with AI :)

Thanks :) I’ll post here when I have larger updates.

I had it starred! It looks like a nice set of tools for building a multi-modal AI app - I'll give it a try when I flesh out a discord AI gaming app I was working on. Is the multi-modal aspect (image, audio, language) the main focus? Maybe putting that a bit higher in the readme would help it stick - the intro section was a bit too dense and I ended up skimming it.

As for traction, I wonder if there just isn't much interest in AI with JS/TS right now, for whatever reason?

Yes, good point, JS/TS is definitely behind Python. That might explain some of it.

I expect most models to become multi modal in the future and am building towards. A lot of the core logic of agents will nevertheless be text based imo, so that’s a central piece, but I already added text to image and speech to text, and plan to add text to speech next.

I've been working on TurboScribe (https://turboscribe.ai/) for a few weeks and preparing to "officially" launch soon.

I think high-quality AI transcription should be a lot cheaper.

It's pretty simple: Unlimited transcription (powered by Whisper) for $10/month. Upload your audio/video files, get your transcripts (and export to many common formats).

There's a free tier of 4 transcripts (i.e. up to 2 hours) per day if you'd like to check it out.

TurboScribe is especially designed for individuals who need to transcribe larger volumes of audio/video (i.e. dozens or hundreds of hours, including at the highest Whisper large_v2 quality level) or just don't want to have to worry about getting billed by the hour, hitting some ridiculously low cap, etc.

Will you be HIPAA-compliant?

What a wonderful idea, would love to have this be a monthly thread.

Homebrew clubs style—bring the half-baked experiments you've been tinkering with, we'll all learn something from it

I agree. This is what HN is all about to me.

I built Carve, a site that lets users schedule Google Calendar meetings with 1-click. There are a bunch of features I want to add but I want to make sure that this is a real problem first.

Link: https://carvemeetings.com/

I currently use Calendly and will try Google's new built-in offering[1].

What features would you add that set Carve apart?

[1] https://workspace.google.com/resources/appointment-schedulin...

I just read through that page, thanks for linking it! My vision for Carve is that I just want users to say "I want to schedule a meeting with A, B, and C people". And then Carve would handle the rest: finding availability, making sure the meeting time is during working hours, etc...

The additional features would let users set filters on:

* What times they wanted to schedule meetings

* Schedule meetings at least 24 hours in advance

* Set some people as optional

* etc..

There’s something along that line here but I don’t know if it’s the best execution of the idea

https://support.google.com/chat/answer/7655908?hl=en

Schedule a meeting with cynthia@yourdomain.com

Find 45 minutes with Cynthia next week, title it "1:1”

These two flows are the ones I'm interested in building out. Something low friction where you don't have to spend time trying to find a common free slot on each other's calendar

https://www.webtospeech.app

It takes a URL and a voice selection, sends it off to Azure TTS speech service. The backend handles converting the HTML to SSML, it's really a one step solution to converting web pages to audio files that you can download, or stream on the platform.

I use it a lot for Hackernews articles at work, where I want to keep working, but there's an interesting article that I want to read.

The voices that come from Azure are reasonably good enough.

It's just an Astro app for the frontend, with a golang service behind it that parses the webpages and sends it off the azure. Also using supabase for authentication and database storage, although I'm using Cloudflare for it's R2 storage of audio.

I'm creating a 2D ant colony simulation that runs 24/7 in your browser. It's supposed to feel like a pet and you're supposed to want to keep it alive and check in on it as tunnels are dug, ants are born, food is found, etc.

You, the proud owner of some new ants, must keep them alive. For the MVP, you'll just get a ration of food each day when you show up, but there's a catch. To unlock the food, you need to arrive at a self-designated time each day, you can't miss your time window, and you'll need to participate in a breathing exercise. Black screen, white text, breathe in/out, repeat for a minute, etc. You emerge calmer than before, get your food, feed your ants, watch them thrive, and get on with your day.

Long-term, I'd like to continually increase the complexity of the simulation to make it more engaging, disguise the self-care offerings behind game lore/mechanics, and provide more self-care offerings such as journaling or guided meditation.

The goal is to get people who like using technology a lot to unintentionally dedicate a few minutes each day to checking in on themselves. They get into it thinking they're caring for some wannabe-Tamagotchi ants, but they stick with it because they're on a seven day self-care streak, feel good about that, and don't want to let their ants down. It's similar to https://finchcare.com/ but less on-the-nose about feelings and more for those who'd appreciate a RimWorld / Dwarf Fortress simulation aesthetic.

This is my first attempt at anything like a ~game and I have very little to show in way of functionality, but I've been enjoying teaching myself Rust/Bevy and writing with an ECS-first mindset. Trying to use these tools has slowed down development several magnitudes compared to throwing TypeScript at the wall, and I haven't seen significant benefits yet, but I'm hoping that long-term the low-level control and fast-growing game engine will pay dividends.

Anyway, here's a few ants which dig dirt and make some sand piles. I know it's not much :)

https://ant.care/

I'm hoping to get people checking back on their ants for seven straight days by spicing up the visuals a little bit, adding a queen, nesting, egg laying, hunger, death, and then a top-down view for food discovery / pheromone trails.

This sounds like an awesome project but a terrible product, but in a good way. You are not hooking the user, but you are not hooking people with gameplay. However, if you can give it a nice aesthetic, this might work as a game on the desktop/mobile.

Regarding Bevy, my wife and I recently started a games company recently and she did a ton of research on game engines. The verdict in Bevy was that it needs to mature a lot more in comparison to more mainstream engines. YMMV. If you run into issues, take a look at Godot. It's has been fairly simple to learn, its language borrows from TypeScript and Python a lot. You can also use gdextension to write code in C++, or even Rust with the community-supplied bindings.

Good luck!

:) Yeah, the "awesome project, terrible product" idea you're expressing has crossed my mind many times. It's an interesting line to try and walk to be sure.

Care to share more on the research you did? I'd love to hear more.

While I'm not a stranger server-side languages, the last decade of my career has been mostly building CRUD apps with JavaScript. So, I initially tried building out this simulation using React + PixiJS. I ran into performance issues with React's reconciler pretty quickly. That, plus no real modern, 2D JavaScript game engine (BabylonJS for 3D .. PhaserJS for 2D?) made me look elsewhere.

My intention is to build a web-first game not a steam-first game because I see focused interaction with the game being just a couple of minutes each day, but brief check-ups on the game state to occur repeatedly throughout the day. I don't think those use cases align with how I see myself using Steam. Usually Steam is a sit down and game for an hour sort of thing.

Assuming web-first is a priority, but avoiding JS, the only option is WASM. I did some reading around and saw suggestions to avoid using GC'ed languages because JavaScript's GC is one mark against it already and because stuffing a GC into a WASM build makes it very heavy.

Those decisions sort of naturally pushed me into Rust/Bevy for better or worse!

The whole games development has been a wild ride so far, we spent 9 months learning it without bringing out any game. We threw away a lot of prototypes though. Our background is in open source and server side stuff, so we needed to let go of a lot of concepts to be efficient.

Research-wise we did a whole lot of reading and a bunch of work building games we threw away. Here's what we found and I can share publicly:

- Unreal Engine: pain and suffering for awesome results. You must love debugging. You get source code access to help with that. Have a Rider subscription to lessen the pain of C++. (I'm not joking.) Loves to crash a LOT, but at least you get a stack trace. C++ has garbage collection here, which is neat. Did I mention to get Rider? Not web friendly though.

- Unity: works, relatively easy to get good results, lots of tutorials, but no source code access unless you fork out a lot of cash up front. Didn't crash on us.

- Godot: Free, open source, easy to start, stupidly hard to master. Needs a lot of fine tuning to make look good. Random weird bugs pop up with no stack trace. Has a custom DSL called GDScript which has some weiiiiird design decisions. (E.g. a Vector2 is not an object, but a data type.) GDExtension takes some experience to set up. I like C++ with UE a lot more.

- Phaser: It's a toy, lacks too much to do anything serious with. Love it though. Might be wrong on this one, only built one small game.

- BabylonJS: Built our website with it (see my bio), sturdy workhorse but you need to optimize your stuff a lot beforehand. Love it. Forums/community are awesome.

The ones we didn't test:

- Three.js: didn't try, but the API is apparently not very stable.

- Bevy: The community is very small, lots of parts seem to be still missing to build larger stuff. Didn't try it out in the end.

- OGRE3D: Couldn't get it to build, but it looks very robust. Would love to try again some day.

- PyGame: Good to learn the basics, but like Phaser, it seems to lack a lot to build more serious projects with it.

Keep in mind, opinions can be very subjective, it can be very hit and miss. We had fans of either comment on our YouTube videos and we have a feeling that getting into game engine discussions is similar to JS framework discussions.

One thing we learned for ourselves though, since we are trying to make money: stop wasting time. There are 20 year olds out there who release on Steam for a chance to become successful while we spent lots of months learning stuff. Technical prowess does not a good game make. It's all about the gameplay and the vibes.

Regarding Steam being a sit-down thing: both my wife and son love the Steam Deck and they often jump into a few short rounds of something. (Brotato is pretty good for that.) It is what you make of it, but the audience does seem to expect a gameplay loop that works over longer periods of time indeed

I hope this helps. ;)

It reminds me a bit of Viridi: https://store.steampowered.com/app/375950/Viridi/

I don't have any suggestions, but I want to build my own version now. Emergent behaviour FTW.

Yes, absolutely! I 'played' Viridi for quite a while many years ago. I always looked forward to engaging with it - even though it was just for a second before switching to a classic game. The creators were also pretty open in discussing their development and business financials which I appreciated - http://www.icewatergames.com/blog

If you end up building something, I'd love to hear about it :)

Call Stacking is a Ruby on Rails debugger that captures everything for a given request - each method call, params/argument/return values, class name/method/line of execution. You can visibly see which method called which, as each method call is nested relative to its parent callee. In a single consolidated timeline.

https://callstacking.com/

The state of debugging is kind of in the dark ages. I bet I spend 90% of the time on a feature just trying to figure out the execution path - what method calls what other method and under what context. It only gets harder as the code base grows.

Call Stacking is an attempt to automate that discovery process.

https://www.inouts.com

I build Inouts - a search engine for online tools without BS, with a focus on specific examples of their use.

Instead of browsing seo optimized pages, fake "top X tools for ..." rankings or websites full of dark patterns where, for example, only at the end of the process you find out that you have to pay, you can use Inouts.

Currently in the database there are more than 5000+ tools and even more examples of their use. Categorized in over 1000+ tasks they solve. You can easily find the right tool through the search engine and filters such as:

- inputs, outputs

- tasks

- payment plans (free, paid)

- platforms (website, mobile app, etc.)

- features (no account required, no CC required, no watermark, etc.)

Currently works best on desktop.

This is a great project. I have a couple hundred bookmarks categorized and sorted for this very purpose, especially for css/html tooling.

Are you manually entering all this data? How are you preventing spam/duplicated content?

Also, some feedback: https://www.inouts.com/explore?query=3d+model

I was hoping to see if there were any 3d model assisted ai tools in the db, but the results aren't highly relevant for this search

Love directories like this!

Great idea and all, but I gotta admit: My first reaction on reading your post was, “Name sounds slightly pornographic.“

Ive started 3 projects i would like to come back someday:

1) HoM&M3: Hota player vs AI battle simulator - i already did some reverse engineering of AI used in battle and reimplemented some game mechanics to be able to run simulations with a few units. Purpose: to find average best possible output for given fight scenario.

2) Benchmark runner, database and visualization - i have some code for db, cli and dsl used for querying data. Purpose: to find performance regressions in subsequent builds and to give nice visualizations of product performance for management and marketing.

3) General purpose scripting language built in Go. No code, but most of the grammar already defined. Purpose: mostly educational.

Over 2 months I monitored block-list data for the shared IP pools that Marketing Automation vendors offer their customers (Act-On,Campaign Monitor,Eloqua,Hubspot,MailChimp,Marketo,Pardot). Now that I have the data, I'm analyzing it and writing an article. I'll be sharing figures for average MTBF, MTTR and the 2 main approaches vendors use to manage their IP pool. If you can think of other insights that people would be interested in, or are interested in doing an initial reading, then do reply!

I am building a Web UI called paisa[1] to visualize personal finance data. We currently have very good command line tooling for this (ledger/hledger/beancount). A Web UI will make it even easier to see what's going on with your finances.

I am interested in knowing what are the common problems you face with command line approach that can be solved via a Web UI

[1]: https://github.com/ananthakumaran/paisa

[2]: https://paisa-demo.ananthakumaran.in/

I have literally so many not done projects (maybe I have adhd)

- TripSync: A “shared travel plans app” for you and your friends. Essentially tells you where your friends are going to be when and when you’ll be in the same area. (For iOS, not yet launched of course)

- cookmatic: recipe organization and parsing: https://cookmatic.co

- polyfill: finding developer jobs based on company values as opposed to programming languages: https://polyfill.work

- tweeker: a/b and multi variate testing: https://tweeker.io

- payote: custom stripe invoice data and invoice plugin for your site: https://payote.co

- minotaur: using Spotify data and YouTube to make a free web based music player with a better, simpler ui: https://minotaur.fm

Some are essentially finished, others are barely started.

I’m building https://coldpress.ai - the idea is to be a large library of labelled datasets (both free and paid), so that ML folks can spend less time hunting down data and more time working on the model.

I don’t want to entirely replace the data gathering + annotation workflow for everyone, but I believe that ready-to-go data can reduce the barrier of entry for ML projects, make _some_ people realize that they don’t really need their own data collection, and even help large businesses reduce their dependency on Scale / Appen.

We’ve been building Design/UI development app with first class support for code generation (react + next) - https://www.kubi.design

It’s UI/UX is pretty similar to Figma (you can import from Figma). We do support integration of your own design system, that means that you can use actual code components in design canvas with all interactions and business logic. (Currently we have integrated Ant design and Mantine libraries)

We are looking for beta testers who want to speed up UI development or simply save money on FE development.

I'm working on an application where you share your business idea with an AI which evaluates and ranks ideas. It scores the idea, explains why it scored the way it did, and offers suggestions for improvement.

That's double meta in this thread, I know.

I don't have an MVP yet but I've been testing and tweaking prompts. An early strategy might be to release this as a free open source app and then offer extra business value later. I created an email form in case there's any interest here.

https://tally.so/r/mK5J6K

There's already a `dimeadozen [dot] ai` for that purpose. What other features will your app have?

I didn't know about them but after a quick glance I'm thinking...

- You don't submit your ideas through us; they stay between you and your LLM.

- It will focus on tech founders.

- It will be open source and free with additional paid features for the 1% who may need more.

- It will show you results without asking for more information.

- It will have a much more narrow focus; at front at least.

A webapp that lets you quickly record over existing YT videos (or TikTok), putting your face over the video in a little overlay circle, so you can make commentaries about any existing video, it will support multiple YT videos so you can mix them for whatever reason (e.g. your video of top 5 doom mods that you clipped together 5 different videos), the main focus would be to make it easy, no dragging bars like most video editing software; you record, you click "undo" if you didn't like the record, by default they are saved in the order you recorded

https://www.casinojoblist.com, which I hope is self-explanatory. Updated daily.

Problem is getting any kind of SEO success. The site is functional and designed to be as easy to use as possible, what it doesn't have are any articles, blogs, ads, pricing, or distractions (other than multiple casino-related themes, for a little bit of flair). What's good for users (I hope) seems to suppress any presence in search engines, however.

Maybe instead of linking out directly to the source site from the first click, grab the job posting's content and re-create it on page on your domain, and then link out to the source site from there?

Appreciate the suggestion. Even if it helps SEO it degrades the user experience (extra clicks before getting to the actual job listing) and a lot of resources (already sending a somewhat large payload to the client). The source listings themselves are not in any standard format, initially I did want to enable a job description on hover/click but they are often unreliable in formatting or sequencing or taxonomy, which is problematic. Still worth considering, thank you.

I have made many projects, some of which are incomplete. In the case of Free Hero Mesh (a puzzle game engine, like MESH:Hero but many enhancements; in fact I originally wrote it because I wanted to play Hero Hearts on Linux) maybe it is too late (it is ready for use although improvements may be made in future; some features are incomplete so far). I also had incomplete (and not working yet) such as Pokemon Text Battle (supposed to be a Pokemon battle simulator in C, with a C API (do you have some ideas about how this API could work? I have some ideas but I am not sure that they are very good (I partially wrote the .h file so far), so I want to ensure to improve it before it is implemented); there are some features I had wanted to implement that other battle simulators lack, including a telnet-based implementation), and a currently unnamed operating system design (I have many ideas but no name yet; this includes a capability-based system with proxy capabilities, a hypertext file system with no file names and no directory structures (file streams can contain links to other files though, both to a fixed version and do a varying version are possible), a common file format for most types of data which can even be (text, tables, time series, and many others), integrated command-line with GUI, accessibility and multilingualization, non-Unicode, and many other ideas; I have written some about it on Usenet).

Back before Apollo the client for Reddit announced that it was going to disappeared, i started working on a client with the same dark scheme and ui feel as Apollo but for HN.

I did this as a way to play around with iPod development and swift and i made it pretty far, more than expected, using the hn api powered by fire ase.

I eventually got into an issue with hn stories with too many comments. I need to rethink my approach on how to handle these cases, otherwise the app crashes when it tries to render 800 comments or so.

Overall it was a fun learning experience.

* iPhone development

Damn, I'm most likely late for this.

I've got a project called http://digitapes.app/ - very much not-finished yet, but would be interested in hearing some thoughts.

The basic premise is that it's an easier media archiving / montage tool. Rather than saving everything to google/apple photos, this allows you to "share" to digitapes instead (much like you might "share" to your family members via messenger) - then digitapes puts them all together into a montage reel that you can watch / download / share.

You could also give access to your friends, who they can "share" it to your tape, so the group of you can make a montage of the holiday you all took.

I also wanted to break away from subscription pricing: you buy a tape for X dollars, and it should continue to exist* (maybe it becomes read-only after a little bit, and it has a size / length limit). - this obviously requires more people to buy more tapes, to keep the business going.

I made it with idea of keeping "home movies" for my daughter. My wife and I have video tapes from when we were young, but digital media has a "cheap and plenty" vibe, so I wanted to have a special place that i could easily store and montage photos and videos of my daughter while she grows up.

I've just been too busy with raising her to finish it off, unfortunately. :D

A website making trail data from OpenStreetMap easily browsable: https://trailcatalog.org/

Myself and a friend have been working on this for quite a while. It's frustrating because it has come so far (showing trail data worldwide interactively on a tiny budget is challenging), and yet it is so far from being something super useful like AllTrails due to low data quality and a lack of relevant features (photos, reviews, etc etc.)

Have you looked into following the links from OSM elements to Wikidata items and their associated Wikipedia/WikiVoyage articles, Commons images etc.?

Ahh this is a good idea, my friend in fact has been asking me to add it forever. I've been hesitating since the data seems so sparse but I think you're both right. Thank you for the feedback!

I've been working on a redesign of a small robot platform I designed in grad school ( https://github.com/psychomugs/r0b0 ). We designed the robot as a small robot construction kit for other researchers and roboticists, and is basically a floating head under a fabric exterior. My last research project used the robot as a tabletop motion-controlled telepresence device accessible through a no-downloads mobile browser.

I never had the time to polish the platform so I did a ground-up hardware and software refactor. The hardware was redesigned as an homage to popular snap-fit robot model kits, with runner layouts and instructions ( https://github.com/psychomugs/r0b0/blob/main/docs/assets/bls... ).

I've tried to generalize the software as an `aconnect` for non-robot-specific hardware devices. The software powers this digital back I designed for my >50-year old Leica M2 ( https://www.raspberrypi.com/news/leica-mpi-a-pi-zero-retrofi... ) and enables quick prototyping of physical IO, e.g. using a MIDI keyboard to control motors or a gaming joystick to control a mouse.

It's mostly a passion project to refine the platform into something more usable than how it was left, but I'm open to suggestions or beta testers!

https://github.com/dbnet-io/dbnet

dbNet is a web-based SQL IDE using Go as a backend, and your browser as front-end. I built it because I was unsatisfied with the database clients out there. Alot of them are top-heavy, unintuitive, slow or expensive.

Here are some of the databases it connects to:

Clickhouse Google BigQuery Google BigTable MySQL Oracle Redshift PostgreSQL SQLite SQL Server Snowflake DuckDB

I have been tinkering with this word game for something like 6 years. It is intended for mobile, though it works adequately on desktop. It is a virtually endless word-finding game that can be played solo or taking turns with a friend. I still need to add some game modes and more interaction, but the bones are there:

https://wordwhile.com/

Be gentle, it's running on a budget server.

I like it.

The sfx are really pleasant :)

Thanks! They are the result of a lot experimentation and iteration.

This is great! What I‘d like to see is a progression, for example stage 1 with a big 8x8 grid which shrinks based on some metric. The game end once you cant form any new words. Smth along these lines wouls be cool!

I quite like it, bookmarking it :)

I've been working on a mathematical puzzle-game, sort of like 2048, but with real mathematics.

Rules:

- Combine cells by dragging over them to create a adjacent path, that either multiply or add up to the last cell in the path, like 2*3=6 or 4+5=9. - Swipe in any direction to move the cells in that direction.

The game uses visual color-code to help the user to get an intuitive feel for which numbers can be multiplied together. Most people will get a sense of this without an explanation, but technically the color coding are the fractions of the number. For instance if the target cell is 12, it has two orange(2) and one green(3). In this case, one can look at cells close to the target that has those colors, instead of performing the calculation.

--- I've been working on this for about a year in my spare time. I think that the most important part that is missing now is a nicer UI, as well as to tweak the controls a bit.

I've had a ton of fun with this one, with various over engineered implementations (it's my personal project, so I can can do that if I want), and learned a ton.

Interesting highlights:

- Generating fun, varied and challenging boards. - Using bitmaps to create a custom but memory-space-effective representation of every action taken on a board.

As for a pricing model, I think it will be free, with a _Buy me a coffee_-button or similar.

https://gotally.fly.dev/

I've created "AlbertBro," an AI math tutor chatbot for kids aged 10 to 16. It helps with math homework by solving problems from photos. Available as a Progressive Web App at https://photo2math.com?hn=1

Mobile users can snap a math problem pic, desktop offers drag-drop, paste & file upload. Crop pics for accuracy, AI converts to LaTeX for analysis. The rest is math GPT with a system prompt

1) can it recognize handwriting? 2) can it identify where a child made a mistake when ‘showing their work’ aka error analysis? A tool that guides a kid to their mistakes would probably be more useful than on that solves a problem on their behalf.

1) Yes. Handwriting is one of the option to input an image, there's an image to latex conversion happening under the hood

2) It should but it doesn't seem to be consistent and it often just solves the exercise step by step. I'll try to tweak the system prompt to enforce a step by step feedback loop style

Breath focused meditation gear syncronized with pulse rate - today people breathe based on a count, like 4-7-8 is a popular pattern. 4 count inhale - 7 count hold - 8 count exhale. problem is the pattern doesnt account for metabolic level, so I plan to use pulse sensor to monitor pulse and breathing count will be in terms of pulses. This might create some feedback loop so not sure if it will work.

The gear I have is a cheap heartbeat/pulse monitor integrated with raspberry pi.

I am working on a note taking tool which is like Notion and Apple Notes had a baby. As an engineer, I gotta take a lot of notes for my own personal use. I found apps like Notion way too overkill and expensive, Obsidian required way too much time to set up to get it to be useful and Apple Notes was a bit too simple. So a half way between the both would be incredibly productive for me!

Got the database stuff all set up - now need to get the commands to work...

I'm a big fan of json schemas, but it's hard to know what one is missing so for years I've looked for a json schema based fuzzer for http backends, now I've started building one. Even if the input data passes schema validation, does the backend crash because a case can't be caught in schema?

The basic idea is you upload the jsonschema to a website and get a binary in return. This binary generates data based on that schema and sends it to endpoints as / how / when you want. Set up a VM with your backend and let it run for a week, look at the report to find any crashes, repeat. Maybe with some nice webhooks etc.

The reasons to do this by sending a binary the user run themselves instead of over the network are that fuzzing will generate a _lot_ of requests, with a binary you can have it running on localhost and cut down on time.

Value prop: Find crashes in backend code and help you improve your input validation schemas.

Later I've realized that it can just as easily be used to generate test data, lots of test data too, but that probably requires different marketing etc.

Hoping to have a PoC up before end of year or so if anyone is interested.

I’ve been building an app that lets you convert rough notes into articles that match your writing style. The app is called PenPersona, here’s the landing page: https://beta.penpersona.com and here’s the app https://app.penpersona.com

The idea is to remove friction in the writing process. I’ve always wanted to be more regular with my writing, and I have a long list of topics I want to write about with notes on quite a few. I have content, converting those rough notes into blog posts is a time consuming process. I can get GPT to rewrite the notes but it’ll not preserve my writing style and will stand out as AI generated content even though it’s not.

PenPersona is able to look at samples of your writing, generate a “persona” from those samples, and then apply it to notes to convert them into articles matching the style from those samples.

While working on this I’ve seen many similar apps which gives me reassurance that it’s a real problem, but I’m not sure I want to enter a crowded market and the product feels more like a feature rather than a full product to me.

https://leftwardsplanning.com/

I'm building a planning methodology for tech teams to use once they know what they want to build in an iteration. This is meant to be the missing near-term tactical planning framework, not yet another long-term-roadmap concept.

Happy to help facilitate planning too with anyone who wants to try it in their team but isn't sure how to start.

I've been working on a domain-specific language for UI designers, called Matry.

The language is:

  - platform agnostic
  - component oriented
  - designed to feel intuitive to non-technical users

I've been at the drawing board for a long time, stuck in syntax design. So I don't have a working prototype. But I do have a landing page and some documentation:

https://matry.design/

https://github.com/matry/documentation

My philosophy is this: over the last decade or so, the design world has been attempting to get closer to developers and collaborate more closely. While there's a lot of excitement and energy from the design side to solve this problem, I don't see as much enthusiasm from the dev side. So what I'm trying to do is carve out a space in the software development toolchain specifically for designers - an approach I'm calling "design inclusion".

The real question is how to do this. Most attempts from the design side fall under two categories - making design tooling more programmatic (aka Figma), and introducing no-code or low-code tools. I think both of these fall short in various ways; it wouldn't be practical to list those reasons here. If this gap is going to be bridged, I think the solution has to come from engineers. We have to define an interface for design contribution, and over the years that I've spent thinking about this problem, I've come to the conclusion that a formal language is by far the most versatile and efficient way to create that bridge.

I don't have a link to anything yet, but I'm working on a mac app that's intended to help programmers observe their own productivity in an objective but configurable way, and provide a more cohesive interface for working with independent project management tools. This is particularly for people who struggle with things that their managers tend to assume that everyone has the same capability of managing, like when was the last time you updated a Jira comment thread that directly connects to a GitHub PR review, that directly connects to a Git Branch. You might go and push some code, but forget to re-request review, or forget that someone asked a question in the thread, because you don't have native notifications.

I haven't worked out a coherent UI for this quite yet, but ideally you'd be able to give yourself a notification that a branch you were working on hasn't seen any updates in some time, or that you did commit but haven't pushed, or haven't changed the status of the Jira ticket back into review.

If anyone has these problems, I'd love to hear if anything like this would be useful.

Aisearch.vip

I have given up on it because it belongs to a market that requires being a large, well-known entity (search). So I just use it for myself since I find it extremely helpful, but I am always happy when random users give me tips ans suggestions on how to improve it further. It's not 100% done and it's capped at one free search per day because again, I am not trying to grow it

As someone who has been majorly affected by the Hollywood strikes, I finally dived in and spent the time needed to make mistakes to continually learn how to program.

That said, ive been working on a node.js PDF Text screenplay parser that will ultimately push out a JSON object according to an initiative to standardize all screenplay objects in screenJSON[0].

Here is the github for my project, it’s messy but I’m about 80% there and just need to spend more time on Regex to catch edge cases. Currently working on a next.js front end that maps out the script into a visual format but that’s not ready to be seen yet. Also don’t judge me on the name, it’s not going to stay as that. No AI as of yet, but could use text classifiers to be able to tell the difference between a key prop, and say, a camera movement but I’m trying to get as far as I can without LLM help of the classifications.

https://github.com/mattrichmo/ScreenBaby.

[0] https://screenjson.com.

Have you tried using any LLM for this task? LLMs are text-in text-out systems and excel at transformations of text.

Oh for sure, but since most screenplays are standardized to a point, the idea is to just use Regex to get there, then use a local version of llama 2 to finish up the classifications.

Where the fun with LLMs come in, is after all the screenplays have been parsed and turned into a dataset that is trained on not just the story but also the cinematography aspects, as well since we’ve broken it down to a granular detail of each element of each scene.

I see where you're going with this. Elemental (motif-based) editing instead of timeline and frame based editing?

Timeline and frame based editing is the end result but this more about the elemental creation and from there, editing it into time based scene. Ive spent the last few years in the cinematography department, and most directors and director of photographers will write the scenes on flash cards and move them around and rearrange them because not every story is linear, even though we have to find a way to present every story in a linear form. And from there each scenes requires multiple angles, shots, motivations and things change so much on the fly, that a screenplay becomes a document that becomes quite dense with non-presented information.

So, I suppose the next step to this would be to parse a bunch of screenplays from different formats, into a single readable format and then train an image model on the frames of those movies we also trained the text model with screenplays on to get a cross reference of what is written down vs what is displayed visually. And we can break down the visual shots with camera movements, steadicam, dolly move etc as well as identify key props in the image model (maybe. Sounds expensive) and compare them to key props in the script. I don’t know, I’m spitballing now but a multi-modal Hollywood film producer would be kind of fun but this totally is just starting as a way to standardize the script in a granular form and to code since I’m not out on set.

I see. So would the dialogue be part of the textual description?

What would be a typical unit? A keyframe? A bucket of frames?

Because the elements are temporal and visual, I think some form of “object in time” relation must be encoded in the description?

You’ve given me something to think about and I think you’re right, there is an element of time missing. In fact there is a few parallel time elements missing from this too

For example, when we are shooting there is a rough formula for how long of a day we need to get those scenes. Usually it’s 1 hour per 1 page scene plus an extra 30 mins added on for each character in the scene. But that doesn’t translate to the final product as that information tells us nothing about how long or important a scene should be in the final product.

But it’s also possible I’m getting too ahead of myself here and maybe there’s another object that is created that includes the scene, production and final product objects instead of jamming it all into this object.

Open to anything suggestions you may have.

It's not clear what the idea is here. Which is why all the questions. I suspect, as you said, that a much more complex data structure would be required to encode all the various aspects of production into its constituent elements and the relations between the elements.

I would guess, that the first step would be to establish how the process of production (screenplay, scenes, camera angles, locations etc.) relates to the final product: scene frames. Each scene must have many shared elements as well. That would have to be encoded too.

I’m still actively developing offline and on a temporary domain and using screenshare demos and online surveys to get feedback.

Developing a new approach to the online local business directory portals where we start with the user and their end goal to then suggest businesses and bundle deals

instead of assuming they’ll start with location, category or keywords (plumber, London) and then have to filter based on price and ratings.

The platform encourages local business to cross-promote, cross/upsell and offer bundles with non-competing business.

An example; user wants their pool and garden cleaned and trimmed.

Instead of searching for two businesses and starting two conversations, and payments, the gardener and pool cleaner can come together to offer a Summer clean up.

One click. One payment. Two services. Service is delivered, and we pay out minus a small commission.

It’s more a multivendor e-commerce marketplace focused on local business and cooperation than a listings site.

And already partnered with a B2B2C messaging app that coordinates business and customers via email/SMS - so it doesn’t fall apart once the deal is sold.

I’ll have something to share soon.

I want to integrate automatic code documentation with knowledge graph/wiki software. Think Logseq [^0] or Obsidian but with code docs integrated into it, both from public documentation sites but also internal/private codebases. I've done no work to begin implementing this and I have no idea if anyone would find it helpful or not. For me, frequently the thing that slows me down most is having to break my coding flow and search for docs on the thing I'm writing. There's integrations in my editor for some things (Rust and Go both have pretty great support for this with their LSPs), but often I need more in-depth docs than what is immediately provided. Further, I firmly believe that most of the knowledge about a codebase should live in or next to that codebase. I'd really like to have a knowledge graph that integrates fully into the development cycle.

[0]: https://logseq.com/

I wrote a service (Google Cloud Run as the backend, with Telegram bot as the frontend) to generate readable ePub from URLs and send directly to e-ink readers. It was originally wrote for reMarkable 2 (using reMarkable cloud), I recently added support for Kindle (by using the send-to-kindle emails). The code is at https://github.com/fishy/url2epub and I blogged about the recently added kindle support at https://b.yuxuan.org/url2epub-kindle.

I'm open to suggestions on what other e-ink platforms to add, as long as they have a reasonable cloud API. I'm also looking for a good e-ink platform to move to personally, as it becomes apparent that reMarkable really doesn't want third parties to use their proprietary cloud "API".

Love this askHN post.

I'm currently working on "JobMailer".

When I was looking for a job and applying to 100-200 positions, I got really good at the process. To me it seemed like a somewhat wasted skill set, because I was just looking for one job (unlike, for example, trying to acquire customers).

"JobMailer" rebundles paid subscription features that a job seeker might use only a handful of times over their search period and automates manual application tasks:

- Find good fit companies and jobs

- Reword resume/CV to mirror or match job description and keywords

- Find recruiter and hiring manager email addresses

- Write cover letter and send email to recruiter/hiring manager to get noticed

- Track email opens

- Single page hub for all application statuses (think Mint but for all different company application login pages)

The goal would be to semi-automate and submit ~50 high-quality and tailored applications in the first 30 days for a user.

https://nimbusws.com

I've been working on a managed service provider for smaller clouds -- right now it's focused on Hetzner.

A few services (redis, webpage analytics) have already been released on Nimbus and next is Object Storage next and it's quite a big lift!

The dogfood tastes great so far! I've been able to consolidate a bunch of instances I was running and mange them centrally which is fine (I had a lot of side projects).

I've been having trouble finding the right people to try it out and give me feedback/see if they'd use it regularly, and if you're running on Hetzner (or have avoided it because they didn't have managed services) I'd love to hear from you.

[EDIT] - I should note, there IS a free tier, but a credit card is required. You can launch an ephemeral redis instances for absolutely free, but I do have a CC gate up to help mitigate fraud (and to see if I've built something people actually want!).

Firefly is a programming language that aims to simplify full-stack development:

https://github.com/Ahnfelt/firefly-boot#readme

It comes with IDE support and such, and the language itself is more or less done, but there are so many other things to do before it's ready for serious work :)

I've been writing a series on _Image Processing_ algorithms in WebGL covering a range of topics such as Color Correction, Blend Modes, Thresholding, Dithering, Convolution and Film Grain.

https://maximmcnair.com/webgl-image-processing

I want to learn-in-public making an interpreter for my fellow C students.

I've been working on an incremental brainfuck-and-beyond interpreter. Brainfuck is nice because you can deal with single characters rather than lexing or parsing. Right now it's just a brainfuck interpreter with all the bells and whistles you'd want animation, highlighting the loop you're in, skipping the loop you're in, memory tape visualizer, decent error messages when you write a bad loop. Here's a preview (kinda ugg because everything has to live in one file) https://ellie-app.com/nG99JJ3wt6da1

I'd like to make a bit of a writing series on this: start with BF with one instruction and a single cell tape. Expand to cover everything in BF. Then expand out to a language slightly bigger until we have the idea of scopes being a thing.

Here are the ways I was thinking of expanding the language a bit in order to bring in new concepts:

- Bring in numbers where 10> means >>>>>>>>>>. This means we need to parse more than one character at a time and introduce new ways of failing. Like maybe 5[ is a bad instruction because we don't want to have a loop in there 5 times (or maybe we do?)

- Add variables. I know that sounds antithetical, but I think it's interesting. Programs like "++^var>_var" would mean "increments the value in the current cell twice. Store that value into a variable named 'var'. Move one cell to the right. Put the value stored in 'var' into the current cell" so then we have to keep track of what variables are what. Kinda fun.

- subroutines or even better function calls so that the current scope of variables can change. I have no idea what that would be like.

Anyways, is there anything like this out there already?

https://github.com/justinlloyd/banderschnappen

I am working on a dedicated GPT AI Dungeon Master. Results have been promising so far. Though it has a ways to go before it is even remotely playable outside of "yeah, it'll break if you do that, let me load up the jupyter notebook that can handle that situation."

https://github.com/justinlloyd/retro-chores

A physical chores list with a digital component. Unfortunately it has languished for a year due to circumstances of life (family health issues). Now that things are settling down I can return my attention to it.

And I found a functional 4K 55" TV on the street last night that was being disposed of, so I'm trying to decide what to do with that (apart from watching TV on it). I am thinking another cat toy, since the cats trashed their last one.

I’ve been working on a Jekyll SQLite plugin: https://github.com/captn3m0/jekyll-sqlite.

Jekyll let’s you use CSV/YAML/JSON files as data source, but they’re unwieldy since the schema is fixed.

The plugin is meant to let you reshape your data before you use it within Jekyll. You get complete flexibility of SQLite, to let you join tables, use indexes etc.

What I can’t figure out is how to best do “dynamic” queries, say a query that needs to show the number of reviews on the restaurant page (restaurants is the table). Each of these page will need to run a dynamic query with the page.id variable.

SQLite Injection isn’t a big issue, since this is all trusted input. But I’d like something better than just Templating queries using liquid.

I’m doing some parameterized queries on the global level (see README). Need to find a nice way to do them locally.

Maybe just putting them in metadata would be a good way?

I am building a marketplace focused 100% on Magic: The Gathering, which I have played for 30 years.

Biggest issues are inventory and buyer-side bugs now. I wished I would have shipped a few months ago, got caught up on a feature I didn't need to engage users.

https://manapool.com

Wait, this doesn't exist already?

I'm working on building better product merchandising with AI/ML for online retailers.

The problem: creating unique, engaging product content is expensive and time consuming for retailers. Hiring content writers can cost $50-100 per product. Multiply that by 1000s of products and then add in constant product churn.

With uniquely tuned AI models, combined with a retailers raw product data, I can generate content that improves the buyer's experience and leads to:

- higher conversions

- improved SEO

- standing out from competitors

The app includes:

- the unique LL models

- pipelines to connect to the retailer's catalog

- a feedback mechanism to monitor the products (add to cart, conversions, on page time, etc) and test new content

I did pilot with a large wine retailer. We generated nearly 4,000 product descriptions and saw a near immediate increase in on page time and conversions. It is 90 days out we are now seeing an SEO gain too.

https://intro.wtf

I'm writing a deep learning course called Zero to GPT - https://github.com/VikParuchuri/zero_to_gpt .

It teaches you everything you need to train an LLM, including the basics of deep learning and linear algebra. You learn the theory and the application. It includes written explanations, code, diagrams, and videos.

I believe that learning should be challenging enough to let the concepts sink in, so it's not a course you can just skim. It also isn't a "just type this, trust me" type of course - I think it's important to always know why you're doing something and how it works.

I've written 11 lessons, and I'm up to transformers - only a few more lessons to go. It's been fun to write, but balancing time spent training models with writing the course has been hard. Hopefully I will get to finish it soon.

I am currently developing an open-source (MIT) database that can be directly accessed from the frontend (browser and apps) , that is distributed, and capable to deliver data in real-time. The primary goal is to provide support for any use-case that requires close proximity to users and, most importantly, it is entirely free to use and run by yourself if desired.

If you would like to view it, please visit: https://github.com/mateusfreira/nun-db

Feedback is always welcome, especially if you have a use-case in mind that you believe it may be suitable for but are unsure. I have already utilized it in many of my personal projects and for a few clients with a small number of users, but I am hopeful that it will soon be ready for larger-scale implementation. Comments here, issues on GH or emails are welcome

I know this isn’t a topic that would do well in the typical HN crowd, but, just in case someone out there shares my passion: I am working on “technoetic” projects.

That word, “technoetics”, beautifully encapsulates what I find most interesting: the exploration of consciousness through technological means.

My ambition would be to merge “maker culture” with fringe topics related to parapsychology, anomalous cognition (remote viewing), mind-matter interaction, … in a kind of open-minded and curious yet self-aware, artful, playful, and not-so-serious way.

To make it a little more specific but probably still very cryptic to anyone not familiar with this: I am working on software to support various “psi” experiments. For example, I want to make it really easy to do timestamped ARV-based predictions of future outcomes.

I don’t have anything to show publicly right now, but feel free to comment or reach out via mail (see profile) if this piques your interest too.

Had a tough year and found coding very therapeutic (not my day job - so go easy!). The results:

https://www.sheetmonitor.com/ - Time-based-analysis on aggregated google sheet data. You can point it at a google sheet and define what data you want to track. It then grabs aggregate data from the sheet at regular intervals so you can see how the sheet is changing over time. Its something thats not possible to do within the sheets themselves for a lot of circumstances. Nodejs, sheets api & a time series DB.

https://dafr.ink/ - Simple Goal orientated project tracking. Wanted to create a tool that was really really fast, sharable without anyone having to login, but with optional lazy authentication. Svelte & cloudflare functions.

KeenWrite is my free, open-source, cross-platform desktop Markdown editor that can produce beautifully typeset PDFs. I started working on it years ago to help write a novel having a complex timeline and I couldn't find a text editor that allowed me to integrate a character sheet with the story itself.

https://github.com/DaveJarvis/keenwrite

Here's what I mean by using variables from the character sheet directly:

https://www.youtube.com/watch?v=CFCqe3A5dFg

Tutorials:

https://www.youtube.com/playlist?list=PLB-WIt1cZYLm1MMx2FBG9...

CommonMark doesn't propose a standard for bibliographic references. I'm wondering if anyone would find cross-references and citations in Markdown to be useful?

Working on https://kindbuds.ai - a platform where you can craft and deploy custom AI chatbots. You can use them on your site or plug them into our network (kind of like upwork for ai) of bots for users to discover/interact with.

Like bots trained on a market: https://www.kindbuds.ai/ght/telluride

Or bots for utility: https://www.kindbuds.ai/ght/spence

Or bots for business: https://www.kindbuds.ai/ght/knowmad

Much is mid development - ultimately I'm deep working on finishing out the sites UX, bot output and functionality.

As someone who enjoys exploring and learning about new fields, I often lack a curriculum of learning materials. To solve that problem, I'm parsing wikipedia to create a knowledge graph composed of "concepts" linked to each other. Every "concept" has some prerequisite of other "concepts", say, in order to learn differential equation I need to first learn about calculus, etc. Therefore, if I were to learn about self-driving cars, I can mark the starting point (concepts I already know of) and the target (most important 50 concepts for self-driving car), and the graph would give me the best path to achieve my target. Therefore, I can be self-guided in my learning with the help of a knowledge graph similar to technology tree in games. Let me know what you think!

I've been working on a new newsletter for the last month and a half called Rooms where I feature creative individuals homes as well as a brief interview regarding their thoughts on design:

https://rooms.substack.com/

Would love any general feedback or ideas!

I know I'm not the target audience, but I can't figure out who it would be. Unless you're intensely interested in interior design,this is a topic one would get bored of relatively quickly.

Great photos. Interior design magazines/articles are a little about voyeurism - try to bring in some details about people's family situation/life drama/get them to talk about an item that has personal meaning for them/talk about how much $$$ certain things cost.

W nutrition planning app that focuses on looking at an entire week, instead of daily goals. It's based on a spreadsheet-system I've been using for a while: the idea is that you can have a pizza night and eat a little bit more on Friday by recovering from the extra calories throughout the rest of the week.

Mind sharing a link please?

I don't have one yet, but will ping you here when I do.

I have been working on a product which uses AI in the backend to improve email and growth marketing.

Our product flow starts with an email validator - The user can load the list of email IDs or, through our integration of famous CRM such as Salesforce or Hubspot. Our flow first will filter out all the invalid mail ids - We're using an ML model and a basic logical approach in the backend.

Once this is done, we allow them to store valid contact data in our product. Once that is done, they can do email marketing from our product. As of now, we're working on the email marketing part.

The plan is to give them multiple templates for them to choose from. By using AI, we are planning to create unlimited templates for them. The messages can be personalised for each contact based on their details and current status.

Email id validator flow is completed, and we're actively working on the email marketing flow.

https://www.semiform.ai - playing around with a replacement for web forms using LLMs to parse free-text responses and create structured data.

Not perfect right now (you can try it out at https://www.semiform.ai/examples to see what I mean) - trying to figure out a few things:

- How to deal with a 'half-completed' response where the user hasn't provided all the requested information? Not really possible for most traditional forms if fields are required. Could be helpful or counterproductive.

- Need to make it simpler for non-technical users to set up an effective form. Right now planning to add templates to help with this, but I think there's more to be done here

I'm writing a open-source library (https://github.com/RaphaelJ/libhum/) and a webapp (https://datethis.app:8080/) that detects the Utility Grid noise emitted by electrical devices (A/C, motors, appliances ...) from Video/Audio recording.

It then tries to date the recording by matching it to historical data.

The signal processing part is getting decent. I'm still trying to collect more reference signals in RU, UK, China and Japan.

Tom Scott made a video about the idea: https://www.youtube.com/watch?v=e0elNU0iOMY

I built a simple little webapp to help encourage taking breaks during work to practice healthy habits, it's called https://takeabreak.health

Eventually I'd like to add auth and a backend, it's using localstorage for now.

https://www.onemodel.org aka OM: available and in daily use, but progress is slow until health improves. A highly effective (desktop) knowledge manager, text UI, Free (AGPL), using Scala and PostgreSQL (working to move to Rust). Something like an efficient combination of mindmap, zettelkasten and GTD in one, with much greater ambitions including for linking instances for collaboration; see site links to screen shots for some idea of how I organize/categorize things. Low-volume announcements email list available (under "community").

I'm working on a mountain biking app to compare trails that I've ridden.

https://trailcompare.com/

Imports data from strava and lets me search and compare multiple rides.

The trail name detection only works in SantaCruz (CA) area.

https://persumi.com/

I spent 3-4 months working on it whilst having a day job. It's a writing/reading/listening platform, kinda like a mix of Medium/Substack/Wordpress/Twitter/Reddit...

I wrote about the tech stack and how I built it here: https://persumi.com/c/persumi/u/fredwu/p/how-i-built-a-mostl...

Still got a long way to go, especially to tweak the user onboarding side of things to explain much better on the concepts.

Would love some feedback!

This is hard for me to grasp. It seems like an automated conversion tool that takes existing posts and then makes them audio. I see it as a tool since your whole landing page is basically an ad for a tool.

Yet you position this as a platform, but don't really show it. There is no sample content nor people to follow. I also don't see typical stuff like how long it will take to read (listen). No leader board. No examples of personas. Why a whole screen to explain the name?

If instead you kept it as a tool but positioned it like AI that first makes you a better writer, second adds emphasis to the speech to make listening interesting and adds music and intros and ads, third preps the file for other platforms.

Your tech stack writeup url is ugly and I would not want to listen to it without knowing some tool had trimmed 80%.

I am working on an application designed to keep runners motivated: https://www.frenzyrun.com

The main features include:

- Capture the flag: Every week, a virtual flag is planted randomly in various cities. Capture these flags to collect virtual coins, encouraging you to discover new areas.

- Virtual medals: Earn medals by running iconic distances. These medals will boost your experience.

- Groups: Compare your performance with that of your co-workers, friends, and family members.

There's still a lot of work to be done, bugs to fix, and features to enhance. However, it has already helped some of my friends and me to stay motivated.

If you are a runner or want to be one, which other feature can help you to stay motivated ?

What cities are supported? You say various cities, so I assume just the top 10 or something

I'm developing this twitter bot to post the uranium stock market prices each hour. After this done, the next step is developing integration with some IA to produce content as well.

https://twitter.com/uraniumstockbot https://github.com/victorabarros/ura-bot

Ahhh I regret not have a landing page buttt...

I'm making a social media app called Spotted. It's aiming to combine the visual messaging aspects of Snapchat and IG Stories with the "opt-in" features of recent apps like Gas and BeReal

Hopefully looking to launch soon!

Homer Draft is fantasy baseball light that Jomboy used to play with his co-hosts on his podcast. I adjusted the rules slightly so that I could play w my family and friends more asynchronously (we just make picks and don't have to draft in order)

https://homerdraft.com

The two week "season" and then break makes it really fun. It gives importance to games that otherwise would just be one of 162. My wife who isn't into baseball now knows all the sluggers, can rank stadiums in order of HRs, and checks weather on game days at ballparks w/o roofs.

Next feature I need to add is separate leagues so other groups can play w friends.

https://kgdev.pl/scanplay/new

It's a NES-tanks inspired browser game where you use your phone as a controller. It's built for local multiplayer sessions and includes a simple level editor (levels are encoded in urls so it's easy to share them). In order to play you have to scan qr code with your phone (for test purposes it's possible to control 1 tank with WASD after pressing "local" button). There's still plenty of scaling and compatibility issues that will probably never be solved but it's been quite fun to work on.

I'm working on an app called ScreenHint. It lets you take a screenshot of a bit of your screen, which then floats there above your other windows until you dismiss it. It's essentially a clone of the sadly-abandoned Snappy, if you're familiar with it. There are a lot of things I wish it could do still, but I'm worried that I'm missing some ease-of-use and polish items still. The app is free and is currently on the mac app store, but I haven't really done much to promote it yet.

https://www.screenhint.com/

This is great - it drives me crazy that the native MacOS screenshot only hovers for 2-3 seconds before disappearing.

Super early stage still, but I’m building a progressive web app that lets you see all of your medical records across your doctors in one app. The goal was to build an app that focused on privacy and offline-first functionality - storing your medical records on your device, not in the cloud. It’s missing a few core features (like backup and sync to a self hosted server, encryption at rest, etc) but it already (should) work for 3000+ healthcare providers in the USA.

https://www.meremedical.co

Costco Companion App, available on iOS and Android stores if you search for it.

Idea is to improve and enhance your Costco warehouse shopping experience, mainly:

- Provide platform to enable shoppers to review and rate products and for others to find said reviews. Costco and Sameday doesn't provide reviews, so we try to fill the gap.

- Recently we started showing very special Costco warehouse deals in our app, there's a certain magic to Costco prices, e.g. if something ends in .97 it is extremely discounted but it's never advertised otherwise so we try to bring more of these special deals to our users.

- We have more features and want to bridge Costco communities together.

But... feel like, something is missing still?

https://qwerki.com/ (ChatGPT powered personal assistant)

Working in a startup of two people means I'm always context switching & lose track of my day. I convinced my brother to build an app that helps us manage our notes & reminders using ChatGPT. We found it to be really helpful for our day to day work & we recently had someone do an awesome review of our app https://www.youtube.com/shorts/x27OhTQ9bA0

Happy to share testflight link for anyone else interested in helping us

I'm working on "is". https://github.com/oalders/is

It allows me to install various things in various environments without getting too deep in to bash scripting.

For example:

$ is cli version tmux gt 3.2 && echo || echo

$ is there go && is there nvim && nvim +:GoUpdateBinaries +qa

$ is known cli version bash 5.2.15

I've also been working on a service to track TLS cert expirations, domain name expirations, server pings and successful HTTP HEAD requests. https://www.prettygoodping.com/

If you play "Highfleet" at all, can you poke around https://hfopt.jodavaho.io and see if you can get it to generate module lists that make sense?

It's an optimizer, just add what you know you want, set some cost / range / speed limits, and it will output the cheapest version of the ship that includes all required modules and has all requested stats.

Some overview posts: https://jodavaho.io/tags/highfleet.html

been working on https://github.com/smol-ai/menubar for a couple months, but nervous to "launch" it because i feel like it hasn't had a significant "wow" moment yet. basically the idea is to 5x the amount of chat output for every unit of human input, to get variety + familiarity with the wide range of chat models out there. its moved from a "smol menubar app" towards its own "ai chat browser". i wonder if i should put int he work to make it a full "ai browser" before i launch it.

Easily generate function metadata (json) automatically for GPT function calling: https://github.com/Jakob-98/openai-functools

My colleague and I have been working on this tool and using it in some internal projects. It works quite well as an intent->functions machine, and makes the process of invoking functions with GPT a lot more bearable. We hope to make it a library which does one thing so great that it will be the industry standard in its niche, so looking for some feedback on the documentation and use.

I've been building Revenut, a web + mobile app (PWA) written in React Native + Typescript that does simple revenue forecasting for a SaaS that uses Stripe.

Stripe's mobile app and others kinda do this already but some of their numbers can be inaccurate (as detailed in the repo's readme) so that made me open-source + solve an issue my own SaaS has with Stripe:

https://github.com/hbcondo/revenut-app

As mentioned in the challenge section, getting data from Stripe is slow so I've been reluctant to put this app in front of others even though it is fully functional.

I'm building https://www.markix.com, a transactional email service that's meant to be boring and reliable.

At this stage, I'm primarily targeting small senders that need great reputation. There are no dedicated IPs. Emails are delivered from a pool of shared IPs that have great time-to-inbox and deliverability.

In contrast to other providers, I do not plan to support click and open tracking to protect recipients privacy. I'd appreciate feedback on that though.

Always happy to chat about email in general if there are any questions.

I am working on a LEGO-like News Feed that allows you to combine different AI models and Filter/Ranking Blocks to create your own For You feeds. You can take control of the algorithm and iteratively train your algorithm on the data you specify. No more black box feeds that try to optimize you into mindless scrolling.

If you have diverse interests (say Tech vs Games vs Traveling), they can each be contained within their own feeds and your recommendation signals will not be polluted between the different feeds.

https://personamo.com

Wasn't satisfied with zotero and mendeley for pdf reference management. Wanted something terminal based and simple, so I wrote a shell script that I eventually rewrote in python. Given a pdf, it reads it, tries to find the article doi and fetches metadata from crossref to rename the file and store metadata in the file itself. I Plan on integrating features like interactive tagging and tag browsing from the terminal.

Any help with code review or additional features are welcome :)

https://github.com/jayghoshter/prem

I'm working on https://openlang.ai (please try it on desktop, there are some bugs on mobile)

It's a web app to learn languages using podcasts, Youtube and chatgtp. I have been using these language learning techniques myself and I know they work better than duolingo or similar apps. I'm biased of course, but that's the reason I'm building this in the first place, I don't think there's any other app as effective out there.

It's not done yet, but it's usable on desktop. Any type of feedback is appreciated.

https://techblogz.onrender.com/

Its is a collection of engineering related blogs articles from technology companies. Its meant for technology professional to discover tech content. It also has a built-in instant search service of searching for a keyword (like 'gpt' or 'kubernetes') across multiple technology company website's blogs with a single search and filtering the results.

Well, I did a show HN a couple of days ago, so technically it might be published, but given how raw state the project is still in, I do not consider it a big mistake to put it here as well:

https://news.ycombinator.com/item?id=37100760

(Hypersprawl/Hypershack is a metaverse/playground in 4 spatial dimensions. where you can upload your 4 dimensional objects and move around and in them. more info in the show HN or https://brainpaingames.com)

I build https://amazing-gains.app where you can manage your workouts and some nutritional metrics. One focus is on data protection, so you should be able to easily download your data whenever you feel like it. It is also supposed to work offline (PWA) as I personally have very poor connectivity in my basement gym :)

Currently it is completely free but I would need to assign certain permissions to new user accounts so they can create exercises and workouts. I would do that for everyone who signs up today.

https://www.bootstrapsite.com

I've been working on a landing page builder for sometime now (mostly just weekends though).

It's written in vue 3 and laravel. The usp is that once you enter the page data you can try on hundreds of designs and style and customise the components without ever touching the data.

My plan is it make it open source but just haven't gotten around to it yet.

Here is one YouTube video I made explaining it

https://youtube.com/watch?v=cN7E6REY7TM

Would love to get some feedback too.

I'm working on pls (https://github.com/dhruvkb/pls/), a prettier and more powerful alternative to ls(1) that adds a lot of customisation and provides a very fluent command-line interface. It aims to be a superset of exa in terms of the features, while being more actively maintained and targeting a smaller subset of pro-users.

It works quite well and is very usable as a daily driver. I'm adding more features to it and making it available to install it via platform-native package managers.

I've been playing around with a limited PCB autorouter for mechanical keyboards. You feed it a KLE layout file, and it spits out a KiCad PCB, layout map for QMK firmware, and various SVG cuts for case machining.

I'm building a publicly-accessible Matrix app [0] for anyone looking to self-host a discord + discussions alternative.

More and more communities are moving to closed-off apps hidden behind auth, and not being indexed by search engines. This is quite unfortunate, and I'm hoping that Shpong can (soon) be a viable alternative that respects the open web.

Shpong instances federate with each other, and I'm working on adding ActivityPub compatibility as well.

[0] - https://shpong.com/shpong/development

I've seen Shpong referenced in couple of articles that made it to the HN front page. There were some interesting ideas there; keen to see how it goes.

Focused on trying to build out a more authentic locally focused travel site, https://www.tipsiti.com/. Ideally positioning it as an alternative to TripAdvisor and the other more mainstream tour bus tourist travel sites. I love Google Maps but how the information is displayed leaves a bit to be desired. More niche sites like On the Grid are phenomenal but somehow don't have the same reach.

https://waggledance.ai another Agent+Data+Tools app, like AutoGPT et. al. I’m trying to apply a nice UX and combine a lot of techniques (namely concurrency, streaming, review, ReAct, loop detection, skill recall) to make Agents more effective. For now I am probably about 85-90% to my MVP.

- working on cleaning up the source for open sourcing

- the first task is failing right now due to a bug

- doc upload is bugged

- tools are disabled

- tons of loose ends in the UI

- Actively developed: database is likely to be wiped frequently until launch, along with any user accounts.

I've been working on a disk cleanup tool tailored specifically for developers. The idea is to help devs reclaim disk space by intelligently identifying and removing old, unused project files, node_modules folders, docker containers, vm images, caches etc.

I'm currently in the design phase and would love to get feedback on the concept, especially from those who've faced disk space issues due to development projects.

Working on & off on a scuba diving site which allows you to log dives, sites, photos & sealife.

Currently only South Australian & Victorian dive sites, but probably the most extensive collection of either.

Live website here: https://divedb.net/

Source here: https://github.com/cetra3/divedb

Currently exploring activitypub integration, being able to log dives and have people follow/comment/react to them from the fediverse

I’ve been recording voice notes to work through problems on walks for about a year and I love it.

There are other transcription apps out there, but they are all focused on b2b, so I have been making one for individuals.

https://www.glowvoicenotes.com/

It’s still pretty early days(first version should be available in about a week) but I would love to find anyone out there interested in using voice as part of their work flow, there are a few different directions it could go in terms of features.

I'm building a few things, the two I'd love feedback on are:

- https://mousetrack.co.uk - a Disney World price tracker that tracks how the hotel/ticket prices are changing over time and which also gives some great tips on how to save money on holidays

- https://chirper.picheta.me - A Chrome browser extension which puts the Mastodon timeline into twitter.com

I'm working on Knowii [0], an Open Source (AGPL) Knowledge Management platform for communities. In the first version, communities will be able to curate/capture/share collections of useful/interesting resources. A browser extension will make it easy to capture information/links.

Later on, we'll add job boards, ways to share summaries/notes, news, vote, tag, explore through user profiles, etc.

[0] https://github.com/knowii-oss/knowii

That summary mostly makes me think of a wiki; How are you trying to differentiate from the existing FOSS ones?

The UI will be structured, free form text will be limited to summaries one can attach to a shared resource.

Focus will be on people and their contributions.

Currently working on a site that allows you to use your data from run/bike apps to visualize your data on a map, make it look neat, and order a print.

I have an early version of it hosted on https://ridemapper.henrygrant.dev and add features as I think of them/when I have time.

https://github.com/henrygrant/ridemapper-svelte

I've been wanting an app that would let me see all my hikes on a single map. I'd imagine my use case is pretty obscure, but I walk a lot, and I'd like to walk literally every street and alley in my city.

I’ve been working on a tool called Phases! (bit.ly/phases-demo). It’s basically a way to expose localhost to the web with a neat UI that let’s others edit your underlying webpage via AI (currently works with React).

Currently working on building a simpler version first that lets you expose multiple ports (like for your frontend and backend), generates a public URL, and connects the ports by rewriting requests in the browser to the public URLs.

Basically ngrok + more powerful deploy previews. Curious what y’all think!

Kind of an open-source clone of Firebase. It's not only a database but also can be used to serve as a REST backend for your frontend.

I've started building this a while ago but gave up around two years ago: https://gitlab.com/restbase/restbase

I'm building Recall Kitchen which is an aggregator to notify consumers of products they use which have been recalled. Right now I'm focusing on the effort to help users tailor notifications so they're specific and actionable. It doesn't help to be told your brand of taco shells was recalled three states over and I'm working to fine-tune the signal.

https://recallkitchen.com/

Ooh, I had a similar idea (but will not be building it out) - let me know what kind of feedback you'd be interested in getting. FYI the Canadian government has a nice central page with recalls. Sometimes a recall comes with cash-back (after you've already gotten all the use out of the product), so that's a possible benefit to highlight on your site. Here's an example that got us ~$150 back https://recalls-rappels.canada.ca/en/alert-recall/expanded-r...

https://catnip.vip

Catnip is an RSS web app with two main differentiators: - it's an RSS 'aggregator', not a reader mode. I believe the browser itself is the optimal reader. Every RSS app has a bespoke reader mode that tends to suck in comparison to the browser itself. - it has a frontpage that's an aggregate of some popular feeds, making it useful without an account. I plan on adding some boosting of popular feeds once I have users.

Not sure it's as early as you are looking for, I've been at it for 10 months, but https://grugnotes.com is my take on a simpler roam/obsidian/yet another notetaking app. Casual ambition to make it a saas business. But of late, just building without too much polishing, trying to get it to a point I can use it effectively in my small business. Always so much more to making something real than you ever imagine.

https://spheria.ai

Currently working on Spheria - a Personal AI that learns from your data to be able to speak for you.

For a long time, I've been thinking of alternate ways to make my knowledge searchable by anyone at any time and forever (docs, posts, links, books, libraries...). Thought that putting a chat interface that acts as a safe guardrail would make the experience nicer.

I'm creating an app that helps teenagers and adults glow up.

Every day, you can send AI your selfies, and it will provide suggestions for your outfit and how to enhance your appearance. Sometimes, people feel hesitant to ask their friends for advice.

Here is the landing page I created with Carrd: https://glowbuddy.carrd.co/

The app is still in the idea stage, but I'd be happy to answer any questions!

Slack and discord bots for QA over docs. It’s the hello world of LLMs, but my focus is on making them polished and easy to deploy into a docker container. Use them for my current company with ok success. They actually out perform some of the funded startups in the space, but it’s an arms race and I don’t think I’ll be able to keep up against the saas offerings.

https://github.com/shelby-as-a/shelby-as-a-service

https://Rarsy.com is an AI Carousel generator for LinkedIn. This is a high-demand application. The site already gets tons of organic traffic daily and has 500+ email subscribers.

Could monetize this eventually fairly easily with a premium version with expanded suite of features and improve the UX.

If anyone wants to help me take on some of the more technical aspects.

--> I need a clean HTML/CSS to PDF export functionality, but it's proving challenging.

Any interest or feedback?

https://blitzbear.com/

Build user dashboards into your B2B SaaS product in five minutes, not five weeks

I'm building software to transparently connect your program or machine to remote filesystems. I'm initially supporting C++ and Go, and later plan to support Python and Javascript as well.

There are a lot of exciting filesystems out there - what I'm focusing on is ease of use and getting a "magic" experience for the end user. Think NFS, but easy to setup.

If anybody is interested in learning more or helping write some language specific implementations I'd love to talk.

I started a CMS bundle for Symfony projects, and released it. It still needs more work, and people to actually use it. https://github.com/SeleneSoftware/seleneCMSBundle

And a skeleton application that uses the bundle. https://github.com/SeleneSoftware/SeleneCMS

Issues and Pull Requests welcome.

Over the first part of this year I built a small CLI tool to help me keep track of and execute bash scripts, commands, aliases, etc. It's not what I'd call "done yet" but not "just started it" either.

https://bashvault.app/

Depending on what feedback/success I have with it, I plan on adding support for more shells, operating systems, etc.

* ROSboard -- visualize ROS topics in the web browser -- https://github.com/dheera/rosboard

Lots of things that could be done with this but a lot of robotics teams and projects got laid in the past year and it doesn't have as many users as it once did. There is also Foxglove which is a full-blown funded startup working on something similar and the more progress they make, the less motivation I have to make my FOSS project better, as a one-person team.

* Air quality monitor -- https://dheera.net/projects/airmonitor/

Unfinished due to the PM2.5 sensor in it (PMSA003I) always reporting 0 and Adafruit not caring.

* BotParty -- telepresence robots -- https://dheera.net/projects/botparty/

Unfinished since COVID isolation ended and telepresence parties aren't needed anymore

* Luxo -- jumping Pixar lamp -- https://dheera.net/projects/luxo/

Unfinished due to lack of time but it does scoot and jump.

* Digital 4x5 camera back -- https://dheera.net/projects/4x5/

Unfinished due to issues with CRA optimization on the microlens array on the Pi HQ camera and unavailability of a real HQ camera with an actually large (e.g. 35mm) sensor.

* E-ink picture frame -- https://dheera.net/projects/einkframe/

Pending to be done: (a) Get rid of the unsightly USB power wire, use a battery or super-capacitor and a thin strip of solar panel on top of the frame, and ESP32 deep sleep mode and have it update once every hour or so. (b) Update it to use a diffusion model in the cloud for image generation (c) Add a microphone and a mode where you can yell a prompt at it and it draws it

In reading about your 4x5 camera back, I think the plastic lens mount might be obstructing the light going to the sensor, the angles inside such a wide format system are fairly large. If possible, I'd suggest removing it. As far as I can tell, there's no lens array on the sensor[1][2], just a normal Bayer filter.

You could get a calibration frame by removing the lens with the sensor centered, and a black frame by putting a lens cap on, to help with corrections.

That looks like a really fun project. With the HDR mode of the sensor you should get some awesome images. Good luck!

[1] https://www.sony-semicon.com/files/62/pdf/p-13_IMX477-AACK_F... - (PDF) Flyer

[2] https://www.uctronics.com/download/Image_Sensor/IMX477-DS.pd... - (PDF) Data Sheet

Thanks!

> I think the plastic lens mount might be obstructing the light going to the sensor

It's actually a low-profile, 3D printed thing housing a glass UV/IR cut filter. The housing is flush with the glass filter unlike the stock Pi HQ camera lens holder. Removing the UV/IR cut results in full-spectrum photos that manifest the same color shift problem which is characteristic of CRA optimization (where they deliberately put the microlenses off-center toward the edges of the sensor.

> As far as I can tell, there's no lens array on the sensor[1][2]

I thought there is a lens array on almost all Sony sensors? [2] for example lists a "CRA characteristics of recommended lens" section on page 47, but this is not the full datasheet and only goes upto page 25. I can't find the full datasheet on the public internet unfortunately.

Well, if you want to try something different, here's a modified camera that removes the microlens array along with color filters to make the sensor monochrome.[1]

[1] https://maxmax.com/maincamerapage/uvcameras/raspberry-pi-hq-...

Alternatively, you could tilt the sensor as you move it to compensate. Perhaps moving it on a curved path to match the optical center? I don't know how well that would work with variable focal length lenses, though.

If you've ever seen a ball cutter attachment for a lathe, or one of those parallax eliminating camera mounts, you could step the sensor along a sphere instead of a plane.

I had an issue where I had a million droplets with an app on port 3000 because nginx is hard and I'm not that smart, so I made a thing that lets you add a domain mapped to a port super easy, and the request gets dynamically routed to the correct app on the correct port. It lacks https, but is otherwise super useful.

https://github.com/donuts-are-good/appserve

I'm writing an NNTP server in Go. It's something I do when I have a spare hour or two in the evening - which is why I've been at it for 3 years when a decent Go coder, not me, could bang it out in 3 weeks. I'm doing it from the RFCs and hope to make it a full spec server not a toy one. Intending to put it on Github when done but wouldn't dare let anyone see it now as it is comically bad and incomplete.

I am building a sort of SQL query engine compatible with several databases (bigquery, snowflake, duckdb, clickhouse, spark) in order to add more power to the SQL query language bringing multi dimensional capabilities to SQL and more. The project has been released few months ago and available on GitHub https://github.com/squashql/squashql

Probably too early even for this discussion, but why not: https://github.com/cconvey/cpu-n1

It's a simple (WIP) ISA and corresponding simulator. I want to get experience adding a new backend target for LLVM, and this is will be the target.

The idea is to be similar to other LLVM targets, but different enough that I can't just copy-and-paste existing LLVM code.

https://otiv.dev - Automation tool for teams that use Notion and Slack.

Basically got 0 paying customers right now, but we are using it on all our internal projects - just started running Google ads to see if I can get some conversions. Would love some feedback - especially if you are the target audience - a team that uses Slack and Notion.

I have a project. I have posted it once herenon HN. I have not received any feedback then, it hasn't received much traction.

It is a link aggregation. Can be used as a RSS client, or YouTube front end for subscriptions.

It is intended for light, personal use, therefore it is not much scalable, but supports user management.

https://github.com/rumca-js/Django-link-archive

I'm working on https://apispire.com - APIs as a service (authentication, mail submission etc..) nothing fancy.

I have to finish the front end but I don't like to do it so I'm constantly procrastinating.. Funny is that I wrote the backend in nodejs, later in .Net, and at the end with golang but I never finish the frontend...

https://www.podshorty.com/

Been working on this for 3-4 months. Summarizes YouTube videos and synthesizes podcasts that you can listen to. Also has transcripts so you can follow along. Click on the timestamp to jump ahead.

Summarization target percentage is set to about 50%. So if you want to listen to a Lex Friedman podcast, this can save you half the time.

I am writing a low-code tool for Google Apps Script and Google Sheets.

Developed in pure JavaScript and can work in both desktop and mobile browsers.

It includes a custom front-end framework that completely eliminates the need for front-end coding.

The tool also includes many features such as a shopping cart, signatures, and fully flexible forms.

Development with this tool does not require a graphical interface, so ChatGPT can handle most of the development.

Vanna: A Python package to generate SQL using LLMs

Package: https://github.com/vanna-ai/vanna

A paper we're just finished: https://github.com/vanna-ai/vanna/blob/main/papers/ai-sql-ac...

https://todotxt.vercel.app. It's a web app implementing the todo.txt format (see http://todotxt.org/). It's an exercise to learn frontend currently, I doubt I could successfully monetize it. Would appreciate any feedback!

AIPL is an "Array-Inspired Pipeline Language", a tiny DSL in Python to make it easier to explore and experiment with AI pipelines.

https://github.com/saulpw/aipl

When you want to run some prompts through an LLM over a dataset, with some preprocessing and/or chaining prompts together, AIPL makes it much easier than writing a Python script.

https://eatfeed.app

If you are like me you have a list of links to recipes on different websites. Maintaining this becomes a hassles and makes it hard to plan your week.

This website scrapes the recipes and saves them in a database, ensuring you aren’t reliant on the original website anymore. Useful if they decide to delete the recipe or disappear altogether.

Hey, just tried to sign up and looks like your email confirmation is hardcoded to `localhost:3000` :)

https://inbox.video

I've just deployed the other day with a very basic design, but basic functionality tested - a way to charge people to send you messages. Trying to onboard first users now

https://www.peoplecast.io/

Internal company podcast platform

Haven’t done much with it yet, but I was thinking of making some kind of surf/extreme sport social network or something. Or maybe an app to schedule carpools to go snowboard and whatnot. Haven’t put a ton of work into the idea yet

Very open to feedback and ideas. Contact in my bio. Thanks HN!

https://www.hazysurf.com/

This idea seems like it's 100% about distribution. If I owned an extreme sports rental shop, I would have an incentive to get more people to go out and participate in the activities. If the app was well made, reputable and secure I would consider putting up a sign in my store or whatnot.

People that run equipment rental stores probably have a facebook group or professional association. If you can befriend someone influential in one of those, it might be a good place to get started.

See if you can work with data from Strava. Generate groups of people that do similar activities in similar areas, and then somehow invite them to group rides. Those that show up could be a great start on getting some network effect.

Though to be honest I have no idea what kind of info is available from Strava, but I feel that the more you can do on your end and the less involvement required by users, the further you could get with a greater number of people. Relying on people coming to you to sign up for yet another social network seems like a lot to ask of people.

https://app.secalerts.co

Our newest vulnerability alert service. We aggregate vulnerability data from over 50 sources into one consistent data schema and allow users to write detailed queries and setup alerting. We've even introduced a pretty good local scanner to keep your SBOM up-to-date.

I created a simple MVP and opted out of having a fully polished product. Still looking for my first paying customer (I launched this week).

https://www.thumbsup.to/ - Collect more user feedback

I’m working on a personal monitor mixer - a device musicians use the tailor their own mixes whilst recording, reducing cognitive load on the recording engineer - that integrates natively into modern AoIP deployments. I don’t know much about electronics beyond Ohm’s Law so it’s a massive learning curve. But it’s the most fun I’ve had in a long time.

Low budget embedded systems are what I'm all about, and I've done lots of audio stuff(And plenty of wondering why audio is still so analog, even though to am extern there's good reasons), always happy to help with something like that!

This sounds great. I had a MOTU AVB setup for a while which does exactly this, but it’s locked to their hardware. A non vendor locked version or app would be insanely useful.

I only thought about this yesterday, so I have not written any line of code yet, but I want to create a DSL for making animations. The animations I have in mind are things like 3b1b, codeaesthetic style videos but generated entirely automatically via the DSL code. Looking for inspiration currently on the language syntax and features etc.

https://www.probarsearch.com/

It's a small tool to sort and filter protein bars.

I often found myself trying to figure out how many protein per dollar or calories I'd get and a friend of mine said he does the same. So we decided to build this in our spare time.

subZero [0] - The TypeScript library to build internal tools fast

In technical terms, it's PostgREST + GoTrue as TS library (written in rust and go, compiled to wasm, with support for Ppostgres/MySql/Sqlite/ClickHouse)

[0] https://getsubzero.today/

I am working on https://calculator2go.com/?utm_source=hackernews - it is very basic. Nothing technical. I would add tool as I feel need. It is a hands-on approach for me to learn marketing, copywriting, etc.

We have been working on https://github.com/ramanlabs-in/hachi for some time. This ia a webapp that lets you seaech your videos and images locally. We really need feedback about ui/ux, features and usefullness of the webapp.

I was wondering how to organise a project, it is only to show minimal code to try things...

* So I was thinking to create branches from 0 to do some work on it.

* Other approach I was thiking is everything in master/main but different folders.

They are meant to be small scripts or structures so I don't see a big reason for having different repositories

I'm working on a todo list that uses 'j' and 'k' to navigate, and a hot bar of tools to manipulate the list. It's a local-first so the data is in your browser. I'll have to sync to a server soon.

https://fruitful.town

I've been working on server provisioning and application deployment tool for a while now, it's my hobby project that I haven't shared much and it is still in the "polishing" state.

https://serverfluent.com

Am working on a 360 degree video player/image viewer mostly pointed at businesses who’d like to integrate the technology without branding or hosting from other providers.

https://staging.sightsurround.com/

Looks pretty cool! As a heads up, as a user the view movement is reversed from what I was expecting. I was expecting for the image to follow my cursor: when I drag to the left, it pans to the right (vertical also should be flipped). The current scheme works decently for following objects that are passing the camera but when it's frozen it feels completely backwards.

https://automatio.co

A no code web automation bot builder. It's in production, but still need to work on scaling the product, users.

The product is almost done, still figuring out the licensing part and starting with marketing/getting some customers in.

It's https://wplytic.com - WordPress Analytics Plugin (data is stored directly in the WP database).

I work on MonsterWriter. The Desktop app is on the market for some time now but now I'm working on a SaaS version.

We are building an open source library for process automation with transformers:

https://OpenAdapt.AI

https://github.com/OpenAdaptAI/OpenAdapt

Feedback welcome!

https://studios.aeonax.com/racers/

Trackmania meets Elite Dangerous but in 2 Dimensions

This my attempt at making a serious game, learning stuff and finally publishing wherever possible.

And I love feedback.

I have this hobby project for personal time tracking I work on from time to time. It's usable enough for my daily use so motivation has dropped since that point.

https://github.com/eliasson/quarter

A websocket API for storing and querying GeoJSON features, it exposes a lot of PostGIS functionality. I built it originally for a client project as part of a data collection app and then wondered if it might useful for others as a service. Thoughts? I worry that it’s likely too niche.

Mapping is definitely not low or no coding enough.

Mapbox has some nice tools but building up the datasources that can feed into it are another barrier.

Happy to learn more

Use kubeswitch to switch kubernetes contexts or namespaces per shell.

https://github.com/spideyz0r/kubesw

It's written in go, I did some releases but still requires more tests.

It would be cool to get some feedback.

I'm still building a low code system with an easy to use component marketplace where you can edit components within the low code tool. Still a work in progress: https://github.com/yazz/yazz

https://minidemo.app

Been working on a way to turn websites into animated demo videos, with clicks and transitions etc.

Hoping that it'll help with producing landing page videos or marketing videos without needing after effects

https://www.bestlms.co

Working on this while working my main role. This is the missing piece I needed for my previous startup so been working to complete the suite.

Would love to connect with any lending/finance guys about this.

https://nap.tv/

I co-founded https://embolt.app to make running membership-based clubs easier -- we just launched our beta!

I’m currently searching around for a handful more clubs to try us out free in exchange for feedback.

New space AI images every 30 minutes: https://cosmictrip.space

I waited too long to publish this but it is still nowhere near done and I just had to publish early.

Missing most features I want to add but hey, it’s live now!

I've been looking for a 'Grafana for personal metrics' (eg. weight, personal finance, etc.). Does anything like this exist? Ideally very simple, stripped-down metrics/observability w/ ability to input data in UI or via API.

There's a fella named karlicoss whose work you may want to look into. See: beepb00p.xyz

You may want to look into https://exist.io/. It's a very indie developer duo out of Australia (IIRC). And also IIRC they were looking for a buyer on Twitter some time ago.

https://territoriez.io/

(No tutorial yet, click on the castle and move around with WASD. E to undo a move.)

Little multiplayer web game made with Elixir and Svelte. Based on the game generals.io.

Not mobile friendly atm.

* Shell script replacement language that will also be useful as a production language.

* Build system that people will actually love. (I hope.)

* Init system that is easier to use than systemd, while also building distributed systems.

* All-in-one VCS, like Fossil, but it can track changes to binary files.

I think you should definitely get some feedback on the details and specifics of the language before going further if you want it to make it to real production use.

Will you have a GitHub equivalent? Git is dominant I think because it's the standard all the tools are built for, no matter how good a competing system is, I'd probably choose on ecosystem. What's the use case list for binary file tracking? That seems pretty interesting!

Yeah, I need feedback on the language. But to do so, people kinda want to be able to try it. I'm still getting it into a demo-ready state.

Since my VCS will be all-in-one, yes, it will have a GitHub equivalent. In fact, it will be more of a Gitea equivalent because anyone could spin one up.

But I'm not worried about beating git (yet) because the use case for binary tracking is assets for movies and games, and movies and games generally use Perforce. Once my VCS has replaced Perforce, I'll think about conquering git.

Do you have a random bullet points lost or example code for what the language should look like?

Taking on perforce seems more reasonable than git at first for sure.

Old version of the example file is at https://git.yzena.com/Yzena/Yc/src/branch/master/src/yao/exa... .

It's old because I had to stop pushing new commits to that repo until I got my new license approved by a lawyer. I have almost 700 commits since then, several of which update the example file.

But the gist of the language is still the same.

My initial thoughts(I'm not a "language guy" so I don't pay much attention to languages outside the billion dollar ones).

* Is it gonna stay AGPL? Will people use it? * I like the integration with shell. Largely because I hate actually coding in shell and this could be a cool replacement. * Overall it looks pretty cool. I'd probably enjoy coding in it more than C * it seems to have a handwritten lexer and parser? * The mutator operator is cool * I'm not the biggest fan of englishlike "push x onto y" * I like the context stacks, I don't quite like having special builtin one with no distinguishing feature or namespace in the name * Function suffixes are neat but a bit too odd for me But I don't think I'd want to use anything "Indie" in production anywhere, and AGPL might scare off people so it never becomes a mainstream corporate backed thing, right?

Seems like you'd have to really carefully make sure it fits the audience who enjoys using indie stuff if you don't plan to get corporate backing. AI free seems like a good thing for that!

Those are really good thoughts! Here are a few answers and thoughts.

* No, it won't stay AGPL. It will be just my license once a lawyer approves that license. And I have actually four licenses; it will start on the most restrictive, but I will move it to the most permissive over time as this repo becomes established and recognized. In essence, I will explicitly follow the process in [1] that tends to happen over time.

* Yay!

* Double yay! Thank you!

* Yes. This is necessary to implement DSL's in it. (My build system will use a DSL.) Also, it will have better error messages.

* Great! This is one thing that Rust is right on: mutation should not be default.

* I'm not the biggest fan of it either; that syntax will change to something better. Maybe just `push <expr>: <stack>`. I don't know. Any suggestions?

* Technically, they are not built-in; they are a part of the standard library, but point taken: there needs to be a better namespace for those standard library ones. (By the way, this means that yes, your own libraries can add context stacks for client code.)

* Function suffixes are weird, yes. I don't know what to do about that. I've waffled a bit between them and having implicit name mangling (as in C++), but that leads to ABI and interoperability issues. But I acknowledge the issue. What can I say? Language design is hard!

And to answer, your claim that you're not a language guy, you are. As a user, you are more right than me as to what is good; actual users matter more than the language designer. If you are a user, you know what's good for you, and it is my job to hit that.

Someone somewhere said that new projects can only spend a few points on "novel" things; maybe I'll have to spend one on function suffixes, or maybe it would be better to spend them somewhere else.

> But I don't think I'd want to use anything "Indie" in production anywhere, and AGPL might scare off people so it never becomes a mainstream corporate backed thing, right?

Well, I am running a business behind it, an actual real business, so it may not be corporate-backed, but it's as close as I can make it. Because I don't want a corporation controlling it.

> Seems like you'd have to really carefully make sure it fits the audience who enjoys using indie stuff if you don't plan to get corporate backing. AI free seems like a good thing for that!

Thank you! That is actually part of my plan (though I do hate LLM's).

Here's my plan: you have people at companies that write shell scripts for small tasks, right? Over time a few of those scripts get passed around, tweaked, and turned into pseudo-blessed company-wide tools.

Well, maybe one of those programmers writes Yao scripts because Yao is just better than shell. If one or more of those scripts becomes a pseudo-blessed tool, then the company now depends on Yao, and they would be wise to pay me a few grand a year for support.

IOW, you nailed my plan: target the indie audience and use it to snare the corporate one. :) You figured all of that out from the language design! You're sharp!

Thank you for the comments!

[1]: https://writing.kemitchell.com/2020/03/07/No-Posse.html

File sharing and soon remote execution over the internet cross OS. Private and no servers.

https://github.com/prettydiff/share-file-systems

https://totalwebtool.com Working on a Chrome plugin to go along with the main website for a convenient way to scan while on a webpage.

I made pdh4engineers.com, because there aren't many continuing education courses for PEs that focus on subjects like software and controls. The site is up and functional, but it needs some SEO and more content.

Here's a design for a better Fediverse:

https://kolektiva.social/@ashwin/110671528116415354

We've been working on TranscribeMe, a bot that transcribes audios to text: https://www.transcribeme.app/en

You forgot to say "transcribes audio to text IN WHATSAPP AND TELEGRAM!" This is really cool!

Instead of forwarding the messages for transcription, can I add the bot to a group conversation and have it automatically transcribe each posted audio recording? Thanks!

I'm working on a web-based text editor called owo, it's not open-source yet, but I plan on open-sourcing it in the near future. I've talked about it on HN a few times before, and people really like the idea.

Wordpress alternative for developers. Built with React.js, and 100% open-source.

https://github.com/elegantframework/elegant-cli

https://signal-bot.readthedocs.io

The actual Pip package isn't published yet, but the documentation should be self-explanatory.

explicable.ai

The frontpage is done, demonstrating the product. ML pipeline is ready. It lacks the CRUD backend though.

-- please review the site before comparing to my goals below, to see if it matches --

It brings insights (with nice pixels) by clustering together wrt the reason an ML model "reasons about" predicting a column in the dataset.

UX is simple: upload a CSV/excel, select a goal column and you are set.

I haven't spent much time marketing it if at all, and have gotten little feedback; and none wanting to pay for it.

If you see some interest in it, I'd love to talk about it. Or even integrate it in another product, for a fee

My own personal mint that supports all of my personal data, finances, yes Apple Savings, nutrition, projections, with offline/multi device sync.

The goal was to avoid aggregators like plaid while avoiding manual entry.

Created a shorts based learning app with community based content ranking https://www.microideation.com/

My project. Built during Covid Time and still building https://emwavetech.com/get-koottali

Making online shopping return reminder app. My wife buys lots of clothes (in different sizes) and returns what doesn’t fit well.

It’s a headache to remember return windows for different merchants.

I have a on hold project called Who is Blocked. It listens to webhooks (GitHub so far) and try to guess which folks are blocked on their tasks so you can support them.

I’m working on a newsletter platform that is not just another social network.

You own the content, I care about distribution.

https://www.kontxt.io

Social web collaboration. Remix the web.

Open to any and all feedback. Thanks.

I'm working on SystemLisp, a lisp dialect for hardware design and verification based on CommonLisp.

https://news.ycombinator.com/item?id=37095477

https://www.bookhead.net/

I'm working on an LLM app which extends GPT-4's code generation abilities to modern libraries :)

friendlymail: an open source, email-based social network. Email a new post to yourself, and the app generates an HTML version and sends to your followers. They can comment and like via email.

I tried a self-hosted blog, but comments and likes were a mess. Email was my solution.

> wait too long to publish something before it is fully polished

what is your idea / project?

I am working on a book. An un-edited draft is available at CodeIsForHumans.com

i want to make my own sqlite driver with go, but i do not know where to start. maybe i can get suggestion from here

Why?

dontdiecollectloot.com

An odd mashup between Diablo, old arcade shooters, with a splash of Vampire Survivors style progression

We've started working on a tool to use CUE to create a single config space across TF, Helm, and other tools in the devops space. We no longer want the pain of HCL or templated Yaml.

https://github.com/hofstadter-io/cuelm (working name)

Our hypothesis is that a unified config space across our industry will bring many benefits.

Our open questions are largely around whether we

(1) replace / alternative

(2) just a skin, still same reconciliation loop from the tools

(3) replace, but support existing modules & charts

https://transcript.fish

I have been working on this podcast transcription project for a couple months and it's been super rewarding.

I listen to a podcast called No Such Thing As A Fish[0], where some researchers talk about their favorite facts they learned that week. Then they riff on it and are generally smart and funny. I listened to the series so many times that I decided I wanted to listen to the show on shuffle, not at the episode level, but at the fact level.

Since I have been playing around with whisper.cpp in python this seemed like a perfect way to combine some technologies I've been wanting to play with.

I ran whisper[1] over the entire podcast and transcribed all the episodes. I had to do this multiple times because I kept messing up. It eventually took like 7 straight days of my M1 processing to get through ~490 episodes using the medium.en model.

4 million words, and an 800Mb SQLite database later, I got the transcriptions done and have put up a nice site for searching through the data.

Now I just need to figure out the rest. Breaking it up into facts. Getting the audio working. Highlighting and linking to words, phrases, etc.

Some cool info about the process so far:

1. The SQLite database is chunked up and stored as static files, and the frontend queries the static files directly using HTTP range requests, so it only downloads a couple hundred kbs when querying.

2. I've been proper using ChatGPT 3.5 free version to help me write python and SQL. It's been pretty game changing as I feel basically no pain from not knowing what I'm doing.

The code is here: https://github.com/noman-land/transcript.fish

Please help if you know how to get whisper speaker diarization working!! I would really appreciate the help.

Also any tips on ways to index[2] or search[3] my database that will be super efficient would be helpful. Indexing matters a lot when querying the database in ranges like this... I have learned...

[0] https://www.nosuchthingasafish.com/

[1] https://github.com/guillaumekln/faster-whisper

[2] https://github.com/noman-land/transcript.fish/blob/maste/db/...

[3] https://github.com/noman-land/transcript.fish/blob/master/sr...

This is fantastic. I'm a huge fan of No Such Thing as a Fish podcast.

Are you part of club fish? I bet the folks in the discord channel would love this.

Oh yeah, I'm in there. Come say hi! I just shared it with them a couple days ago and they already pinned it! Felt pretty great.

Naive comment here but you could check out using a vector database for semantic searches. Check out chromadb.

Thanks for the suggestion. Vector databases have been on my list of things to check out so this is timely.

i've tried a couple and probably prefer Marqo to the rest. definitely the most user-friendly pick of the bunch.

Thanks!

guessthatword.com

Talk with an LLM to guess the secret word it's thinking.

Currently working on an (unofficial) VSCode extension for tldr.sh

Shouldn't take much longer but I haven't released it yet. I have been playing around with how I want the commands to be displayed. Simple but useful for me, and I plan to keep it open source.

I've been working on a specific reverse-engineering technique called unlinking [1] on-and-off for the past 16 months or so. I'm on my third prototype (first a set of Ghidra scripts written in Jython [2], then a fork of Ghidra [3] and now a Ghidra extension [4]) and I've started a blog in order to document it [5], which side-tracked into writing a whole series of articles on reverse-engineering to introduce the topic.

What for, you may ask? Basically I'm trying to decompile a PlayStation 1 video game and I've quickly decided that dealing alone with multiple +500 KiB executables of complete utter spaghetti code wasn't going to work. Instead, I've decided that I'd rather divide-and-conquer the problem, so I've been tooling up to split executables into relocatable object files, in order to decompile those one at a time and Ship of Theseus-style my way to success.

Ironically, all of that stuff is so not done that I don't even know what meaningful feedback there could be. My prototypes do work, but only for 32 bit little endian statically-linked MIPS executables. The articles on my blog are draft-quality. As for the decompilation project itself that started all of this, it hasn't seen much progress due to all of those side-quests. The overall topic is so esoteric that so far I've only managed to hear about one group of two persons that tried to do anything remotely similar and one another anecdotal account [6] suggesting that this particular skill is very uncommon among reverse engineers.

Personally, I'm starting to think that maybe I could've actually reverse-engineered and decompiled the game in the time I took to get here. I've also tried to engage with Ghidra to upstream the foundations of my modifications in my fork, but after some back-and-forth it became clear that my prototype-grade stuff wasn't industrial-grade and couldn't be merged in its current state, which is why I'm currently reworking the code in my fork as a Ghidra extension.

To those that want to provide feedback after reading all of this: beware, I've had a lot of fun going down that rabbit hole, but this is one hell of a time sink and a particularly tricky mind-bender.

[1] I don't actually know what's the actual name for this technique, given that there are so few resources on it out there. I do know I didn't invent it.

[2] https://github.com/boricj/ghidra-unlinker-scripts

[3] https://github.com/boricj/ghidra/tree/feature/elfrelocateble...

[4] https://github.com/boricj/ghidra-unlinker-extension

[5] https://news.ycombinator.com/item?id=36575081#36590078

[6] https://news.ycombinator.com/item?id=35729232&p=3#35740761

Reply on news.ycombinator.com