meitros

Wordpress does one thing much better than headless CMS tools, which is that it empowers the non-technical folks who actually own the website to be able to change all of the site without friction. Headless CMS tools work well for writing content like blog posts and templatized webpages, but say someone in marketing or PMM wants to iterate/refine the text on the homepage - that requires them to have the developer make those changes.

(You could give them GitHub access and let them use the editor, but in practice they might use it for urgent situations but not more regularly than that. They have 50 other important things to do and editing in GitHub is going to feel awkward.)

Is the opt-in versus opt-out point at the end of the article really important? Even if the compiler is opt-in you could use a separate tool to enforce preventing unsafe memory idiom usages unless annotated (akin to linters from other languages), and C++ codebases don't tend to have the same sprawling uses of third-party libraries that other languages do.

This reminds me a lot of the NY Times' Library project: https://github.com/nytimes/library. You use an editing environment that people are familiar with (google docs), and you build organizational and workflow stuff around it. Library rendered the document content itself with a link to edit (favoring the reader use case), whereas Hermes embeds the google docs UI.

The lack of code blocks in google docs makes it tough for a centralized document repository for an engineering org. For companies using Quip it could work really well...except that I don't think quip lets you embed the editor like that.

Everything that's been built so far for Hermes looks cool. My personal opinion is that it'll need more UX iteration for it to really take off.

Was also curious how the comparisons were being done. Isn’t there a new non-negligible overhead of converting and passing data structures between js and rust?

Interestingly I came to the exact opposite conclusion from that quote - when people stop working full-time the loss is greater than just the percentage of time lost because you lose that extra edge from them like shower thoughts. It’s natural when your job becomes less of your main focus.

I never quite understood the relationship between oracle Java and openjdk. Could a company have zero licensing fees if they were to only use openjdk?

There’s a new accessibility course coming soon too https://practical-accessibility.today/

/ If you want to improve your skills with css then https://css-for-js.dev/ is a great course.

/ the above doesn’t help you in figuring out what you want to make something look like, only the ability to implement a design in css. If you want to be able to design different things or understand designs better (e.g. how/why do we choose to round corners), then https://shiftnudge.com/ is a terrific design course.

/ Will also throw out refactoring ui as a course to help developers understand design more, as opposed to shiftnudge which is full bore learn to design.

I think this blog post would be helpful reading: https://charity.wtf/2019/01/04/engineering-management-the-pe...

The fractal of bad design article is always linked, but I'd suggest also taking a look at https://www.youtube.com/watch?v=i5E48THNRe4 (or https://www.slideshare.net/zerutreck/taking-php-seriously-ke... if you prefer slides to videos.) It's by one of the developer's of Facebook's PHP engine/variant (Hack), and it tries to nail down what does make PHP special, beyond all the obvious warts.

(FWIW, one thing Facebook did have in it's favor is that it got pretty disciplined about what parts of the language to use or not use early on (e.g. no extract), which IMO made a large difference compared to typical php codebases back then.)

Bootstrap with python flask will probably be the easiest way to write the ui code, imo. If you then add in one library - htmx - then you can have an idiom where the server handles all the rendering and you can avoid using javascript.

I think these tools should work for someone coming in without prior experience and figuring it out from the docs.

That was definitely the weirdest point of that post. Even if you agree with "hire generalists over specialists", hiring generalists to build a custom S3 sounds like a terrible idea.

If I had to guess, EA is Effective Altruism

no wonder he needs your help to complete levels

This seems like an interesting heuristic for anything that could automatically either generate or "format" code - a little more semantic than just relying on a text parser

Liveblocks looks great, thanks for mentioning! Will try it out sometime

This wasn't mentioned in the article but IMO belongs as the last section of the timeline -- To (one of) the points made in https://news.ycombinator.com/item?id=32959397, the latest wave of Javascript frameworks wants to answer: how can we build apps like Linear or Figma in the browser? https://replicache.dev/ is the most direct example of this, and https://cord.com/ is an example of a service to add multiplayer elements in a non-framework way without rewriting the way your app works. Also know of one that still hasn't launched yet.

IMO this newest wave is creating frameworks that have existed in the mobile app world for a while and are now showing up for productivity-oriented web apps.

Recoil = Jotai (mostly equal.) Check out the comparison doc https://jotai.org/docs/basics/comparison

This was fun! Would definitely play a mobile puzzle game with levels like this

People do this, e.g. building a cost-based optimizer for queries, but it can take over 2 years for it to really get pretty good. And even then it can only go so far and you'll still want to manually tune certain queries.

Not even just an electron Linux app?

“There’s one guy on the [Fuchsia] team who has a history of leaving a company and then spending the time after he’s left the company to transcribe his brain into GitHub. Because that’s the one thing — it’s in his head, and if he puts it on GitHub, it’s his forever.”

Public repositories? Who is this/this would be interesting reading.

How about looking at the early git history of a project? One in particular that I’m thinking of is Excalidraw https://github.com/excalidraw/excalidraw - it started out pretty simple and evolved over time

Some interesting previous discussion from a stripe engineer who worked on webhooks (the hn story is also worth reading)

https://news.ycombinator.com/item?id=27823109

That always surprised me a bit, because part of the initial intent of react native was to be hybrid. Facebook had a lot of php developers that they wanted to enable to be able to write mobile screens. Not the core experiences, which you wanted dedicated mobile folks to write native for performance/transitions, but secondary things like the reporting flow.

But it probably takes a dedicated developer focused team to manage hybrid. And in terms of pure react native…I don’t know if react native is now able to feel as good as a native app, but even for a few years after Airbnb’s post react native would not have been up to the task of rendering all the little animations and transitions that Airbnb had in their mobile app.

(Though these days I would 100% write a new, less fancy mobile app in react native/expo - it’s constantly improved throughout the years)

re CLI, the author of esbuild had an interesting comment about writing it in both Go and Rust and the Go version being faster out of the box https://news.ycombinator.com/item?id=22336284

I wrote an internal tools framework on top of htmx with a desire to make things easier for backend contributors, but ran into exactly this problem - even htmx isn't quite simple. What I would have likely done next is try to create jinja abstractions around htmx

It has the same creator as React as well (Jordan Walke)

Did you delete your profile text? It looked empty to me.

(Cool translated profile name, though :P)

Quill tried to solve this, but they eventually got acquired https://web.archive.org/web/20220122150126/https://quill.cha...