Show HN: How to X with Y. Engineering questions, answered | Hacker News Reader

Show HN: How to X with Y. Engineering questions, answered

howtoxwithy.com

I get it, this is supposed to more usable than Stack Overflow, where pedantic discussions can obstruct usability by occluding the simple Q&A. But right away I see the simplicity of "How to X with Y" yields solutions that are not entirely correct (which is different from incorrect). For example, the current solution for reading a file in Python is unsafe, yet there is no way of seeing how or why; sure it works, but it also proliferates dirty code.

Totally agree with you.

The idea is that you can have multiple examples per language, the highest upvoted one is displayed first, then you can page through the others.

You can also comment on examples.

But yes, care needs to be taken for sure.

Why is the python one unsafe? Lack of with-block?

When I look at the website I currently only see "No items", but if you are using Python 3.6+ the pathlib[0] library has a very nice way of reading/writing files:

  from pathlib import Path
  p = Path("my_file.txt")
  p.write_text("thing that needs to go into a file")

[0]https://docs.python.org/3/library/pathlib.html#pathlib.Path....

It seems up to me https://howtoxwithy.com/how-to/b35668da-7cbe-456f-8db5-0d645...

That's the "Write to a file" item, I can see two Python examples there.

Ah, this morning when I went to the site it only showed "No more items" no matter what I did. Now it shows "No more items" for 2 seconds and then starts showing items.

I think the first example is what another commenter meant as unsafe. My example is under the hood doing exactly what the second (safe) example is doing, but without the need to use the context manager yourself.

Why is this better than stackoverflow? For example, [0] vs [1] (writing to a file with python). If I want to write to a file in python, I do'nt care what options there are in ruby or JS.

Is there a way for me to flag something as incorrect? As an example, in [0] the python implementation should really be the SO link (using `with open() as f` rather than manually closing it)

Finally, and most importantly, there is no mention of the licensing of the code snippets here. StackOverflow [2] makes it excruciatiangly clear what the license is. Anyone who is using this for work purposes shouldn't touch this with a barge pole without a very clear disclaimer on the license of the code.

[0] https://howtoxwithy.com/how-to/b35668da-7cbe-456f-8db5-0d645... [1] https://stackoverflow.com/questions/34204523/how-to-write-to... [2] https://stackoverflow.com/help/licensing

Of course Stack Overflow is great. It has a wealth of information, experts and allows for larger questions to be asked than "How do I write to a file", like, "how would you design this database schema".

The idea is that anyone can submit alternatives, the one with the highest number of votes will be most prominently displayed. I personally am not a Python developer and just seeded some initial snippets, but hopefully someone who is submits new ones.

That's a great point re. the licensing, I will have to look into it.

> Of course Stack Overflow is great. It has a wealth of information, experts and allows for larger questions to be asked than "How do I write to a file", like, "how would you design this database schema".

Googling questions brings me to stack overflow which has years worth of information, and answers to all of these how do I do X with Y questions, along with links to the documentation. Why would I want to use this site rather than stack overflow, what does it do differently? If SO has everything your site has plus more, (and lets be honest, SO has answers for those questions), I don't see the benefit here.

> I personally am not a Python developer and just seeded some initial snippets

While I understand why this is done, this is exactly the issue. The code you have right now is wrong, but stack overflow's is correct.

> Is there a way for me to flag something as incorrect?

THat's the problem with these types of sites (and the problem with SO), most of the answers and submissions are by people who don't know what they are doing.

I think these sort of sites always need a big disclaimer on every page (PLEASE DO NOT DO THIS IN PRODUCTION CODE). It's the blind leading the blind.

I guess it depends on your domain, language, type of question, etc .. but in my experience you get a good amount of correct answers in SO, buy clearly YMMV.

I do find answers that are marked as correct that are no longer the best answer, what i find less of is obviously wrong answers marked as correct

> ... are by people who don't know what they are doing.

I wouldn't agree with that. However, answers might get outdated.

Definitely, I still don't have a good answer for this issue.

Hopefully something reasonably organic arises where good answers are upvoted highest. Or there could be a way to mark things as using an old version.

Just a flag that says "This answer was submitted N years ago - it might still be valid, but please check compatibility etc...."

I'm genuinely curious to hear your suggestion on what to do instead? How can one get a free answer online that is guaranteed to be the actual best solution to the problem?

Just out of curiosity, if SO examples shouldn’t be used and these type of sites should be avoided, where do you suggest someone learn the proper way to do something in x lang?

My understanding was that SO is basically the modern day O’Reilly cookbook.

This comment/poster reminds me of the verge’s pc build fiasco aftermath.

‘ThAts nOt HoW YoU aPpLy TherMal PasSte’

That was just piling on, the dude called a livestrong bracelet “anti-static,” made stuff up about how to install a psu... the list goes on but that was a legit fiasco on TV’s part and their handling of it showed they strayed very far from their original mission statement.

For an amateur like me, who’s more concerned with getting his code to actually work, something like this is actually less daunting than stack overflow.

The problem with stack overflow (again, for someone like me) is that issues posted on there get abstracted to oblivion, resulting in posters proffering a million different, mostly correct, mostly complicated, solutions. Also, I believe OP’s product has a way narrower scope than stack overflow.

SO is not going anywhere anytime soon. I just don’t see why it has to be the end all resource.

Assuming OP can find a good way to deal with duplicate entries, it seems like a more “to the point” version of SO that can be referenced before having to deep dive SO’s massive sea of information.

Just a layman’s two cents.

Agree with all this. I still use Stack Overflow all the time, when I google some strange issue I'm probably not going to land on howtoxwithy.com, but Stack Overflow.

The issues that get generalised into oblivion aren't the sort of issues that will end up on howtoxwithy though. You can't really abstract "concat two strings" any more, and both howtoxwithy and stack overflow both have the same answers for python: x + y

> SO is not going anywhere anytime soon. I just don’t see why it has to be the end all resource.

I don't think SO has to be the end-all resource, but to me it seems that this site is just a subset of SO's functionality, and doesn't provide anything different.

Hey HN,

I started this project last Tuesday to scratch my own itch. It's basically copy/pasteable code snippets for common tasks in many languages.

Like many developers I work in different languages and often forget syntax or how to do simple things. I find if I'm working in a big codebase I can usually get comfortable quickly, but it's when I'm faced with a blank page in my editor I think... err how do I do this again?

I've had this idea but with libraries instead. Neat to see some similar lines of thinking.

Some questions:

- Is the project open-source?

- Why no github signin?

- How do you search for X but only in Y language?

- How are you going to deal with duplication?

- Will you be doing a full 2D spread of combinations? E.g. writing to a file is pretty common - perhaps you can create a relationship between languages so that you can see how it's done in other languages from the same page.

- How will you handle cases where there are multiple ways of doing things?

- Is search broken? "Write to a file" is the first entry there without searching, but searching for "write to a file" shows no entries. - How will you handle cases where environment matters? E.g. writing a file in Javascript is going to be different between web browsers and Node.js, and force-inlining a function is going to be different between clang and msvc, etc.

Cool! Yeah that's another good use case. With ecosystems like NPM (ie. ones that have so many packages) it can be really hard to know which library to use.

- Is the project open-source? Not open source yet but no reason why it can't be.

- Why no github signin? Good point. I recently mulled over which O Auth providers to support and for an MVP I decided Google was enough.

- How do you search for X but only in Y language? - Will you be doing a full 2D spread of combinations? The idea is the page "How to write to a file" has all the language examples

- How are you going to deal with duplication? If it gets popular I will deal with that issue then.

- How will you handle cases where there are multiple ways of doing things? - How will you handle cases where environment matters? You can have multiple examples per language. The one with the most votes will be most prominently displayed.

Sorry, updated my post slower than you answered :D I had a few other questions

No worries! I just updated too. Hope your questions have been answered

My first reaction to the page was: how do I know who wrote these? How do I know whether they are right or wrong? How do I correct ones that I can see are incorrect (or at least misleading)?

I don't see any "about this site", "who created this site", etc. I don't see any information about who authored a particular snippet. Perhaps more information is available if you log in? But it seems like it should be possible to evaluate the trustworthiness of a site _before_ sharing information with it, so I don't feel comfortable interacting with the site as it is.

Good points, thanks for posting.

I stated elsewhere in this thread but the project is only 6 days old. I haven't had time to code up an /about page, reputation system and user profile pages just yet.

While this project is very young and is narrow in scope, I'm really excited by the somewhat parallel idea of building a good enough rosetta stone for practical computing such that one could power a programming platform with human language syntax. Imperative programming and natural writing have an overlap I think is under-appreciated -- the amount of people that could grok computer programming if programs became essays, functions became paragraphs and such is large I reckon.

Has anyone seen anything like this before? This submission seems like it could be a first step, but it's not the authors' goal.

Somewhat more on-topic, learnxinyminutes[0] is fantastic for jogging my memory on languages I use rarely.

[EDIT] - Oh also, I think that this kind of project could have some enterprise viability -- having private instances sharable by tech teams could make it much easier to on-board and share code snippets/best practices with team-members and keep tech-debt knowledge fresh.

[0]: https://learnxinyminutes.com

What you describe is legalese. The precision of programming with the syntax of natural language.

Programmers of legalese are ... special. Even fewer people get into that field than get into programming. Because expressing your thoughts and modeling reality precisely is the hard part. Not the syntax.

But the domains are drastically different. I assume you mean legalese as is used in Law -- modeling reality of law is hard, but modeling reality for computers is not hard, we've worked hard to make things very deterministic and demonstrably so.

All we do is send instructions to computers all day that are mostly executed perfectly, changing the language you use to speak to the computers is the point of higher level languages, and I strongly consider english to be just another frontend (in the LLVM sense), just one that is a lot harder to encode the semantics of fully. But the thing is, you don't need to encode the semantics fully -- just make paths one at a time for certain functionality (ex. `read_file(path: string): string` can have it's first "call sentence" be "read file from <path on disk: var> and store the contents in <new const var>")

Yes modeling computers is easy. But that’s useless.

Most of my job is translating the PMs fuzzy brain and the company’s fuzzy business processes into precisely defined things computers can follow. That’s the hard part and very similar to the challenges legalese faces.

How do you ensure a fuzzy “I want X” happens correctly in every case?

Even something as simple as “90%” can be hard to define in the right context. Does “Read 90% of article” mean you read 90% of all words? Saw 90% of all sentences? Saw 90% of paragraphs in full, or is every paragraph at 90% also good? Etc

learnxinyminutes looks cool! I didn't know about it before starting this.

I agree with the enterprise knowledge sharing! Every company has many snippets and commands they always use

Wasn't it the idea behind visual basic? If I recall correctly the devs aimed at making a more natural programming language.

I don't think it'll ever work, there's the obligatory xkcd comic about the limits of sophisticated programming languages. The code will never free you from the burden of having to specific exactly what you want.

Maybe an ML/AI parser for natural language would be a possible improvement and then tweak the details after the parsing. Likely just make things more complicated though.

> I don't think it'll ever work, there's the obligatory xkcd comic about the limits of sophisticated programming languages. The code will never free you from the burden of having to specific exactly what you want.

In my mind it's less about freeing the user from specificity, but just changing the language under which that specificity happen. Honestly it's just syntax! Imagine something like:

> open file on disk at "/some/path/to/a/file" into $my_file_handle. read contents of file handle $my_file_handle into new variable $contents. print $contents to standard output.

Imagine every sentence just being a textual template for a function invocation, with context in-between functions (state) being preserved and managed implicitly with the use of variable names & prose. signatures for the functions required would be something like:

- open_file (path: string): file_handle

- read_bytes(fh: file_handle): string

- print(s: string): void

We already have functions like this and compose them in ways that the text suggests -- it's more about just adding a different way to group and recall up the composed functionality.

Could you create a github or gitlab, at least to track issues?

For example: When I try to verify my account, (using https://backend.howtoxwithy.com/auth/confirm/BIGHEXNUMBER) I get a message:

No webpage was found for the web address: https://backend.howtoxwithy.com/auth/confirm/BIGHEXNUMBER

Also when I used https://howtoxwithy.com/how-to/615701ba-4707-423d-a012-47943... It let me login as David Whitten but when I used the NEW Button, two things:

1) I can't put in the language name I use. This increases the pain of using an obscure language. Since I use a very niche language, it is frustrating that languages that I know have significantly smaller worldwide code bases are there, but M is not. The page

https://github.com/csurfer/gitlang/blob/master/languages.txt

lists all the languages accepted on github as of 2016 (it is a simple list, suitable for a dropdown list). The page https://github.com/github/linguist/blob/master/lib/linguist/...

gives more details about the languages recognized by linguist.

2) Not reproducible: When I choose a language (like JavaScript) the dropdown list is blank for that item. I expect it is just that the text color is the same as the background, actually)

Hey Dave

Thanks for that report.

I've fixed that issue with the drop down. I'll need to look into the account verification issues. I'm having no problems on my end.

I have a the development/bug tracking board here https://app.portabella.io/projects/daa40d3e-655f-4cb4-b4f4-6..., it's read only for the public right now. I'll get on to using a different list of programming languages shortly

It would be cool to see a solution to some problem written out step by step not in code but in English (or the language of your choice). Then from there link to the code example that solves that bit.

Having to grasp logical problem solving _and_ code syntax at the same time can be overwhelming when just starting.

Can you expand a little on your suggestion? How deep into semantics would you go when describing something?

ie. making an http request, do you include DNS, headers, TLS or is it simpler to just say `https.get(url)`

Off the top of my head, I envisioned trying to teach someone. First, I'd write out the problem. Then I'd break that down, again and again. All in English (read: no code).

Then I'd say: Now that we've dissected the problem, let's write the code.

Yeah. High level. But I can see it making learning simpler as it separates the problem solving, from the implementation (code). And the same problem could be demo'ed across multple languages.

This seems like a really bizarre interface. When I first opened it in Safari, I was shown exactly two examples, and despite the underlined coloring of things like Python and Go as they rotated through the header, they're not clickable.

I'd given up on the site as broken before I saw other comments here suggesting I missed something, so I tried loading it again, this time in Chrome, and there I saw more examples, but still all JS. with no other obvious options or way to find other languages.

Finally I noticed "5 other examples" in what is literally the least-highlighted text on the screen, and that led to me other languages.

Discoverability: 1 out of 10

As the header text rotates, it shifts the rest of the content up and down which is not helpful for reading or targeting.

All in all, I won't be back.

Hey, thanks for the honest feedback.

Looks alright for me in Safari, the snippets shown are just the ones with the most votes, right now those are all JS.

I do agree that the interface isn't 100% intuitive, I'm not much of a designer and I whipped this together in the last week. It will get better.

Thanks for your response. I pooped on your effort pretty directly, and you responded with grace.

Maybe I'll check back in a few weeks after all!

Speaking of the UI, is there any way for it to show something when loaded without JavaScript enabled? Right now I just see a black screen. When I load Stack Overflow I see text and code snippets, so I will probably keep using that for the time-being.

Yeah. Definitely I need to get on the server side rendering train.

It's a React app because I know React and wanted to get it out the door quickly (read 6 days) which is overkill for the functionality provided.

Interesting idea. Just a note that the first answer I saw[0] is not correct.

This is given as how to write to a file in Javascript.

const {fs} = require('fs')

fs.writeFileSync('/tmp/myfile.txt', Buffer.from('hello'));

But you can't just do this in plain old JS. It needs to be Node running ES6+

[0]: https://howtoxwithy.com/how-to/b35668da-7cbe-456f-8db5-0d645...

That's true. Definitely need a way to distinguish between the two (browser vs. server) sometimes

Hmm. I logged on thru Google OAuth, and now I'm logged on as Alex. I've done the List Comprehension Python list filtering here: https://howtoxwithy.com/how-to/615701ba-4707-423d-a012-47943... but that one's written by "Alex", i.e. you.

Seems that if you register with google Oauth the username is always Alex ... or at least for the two of us

That was an oversight on my part. If you sign out and sign back in I can edit the database to give you credit.

Did you know lists have a .remove() ?

    >>> my_array = [1, 2, 3, 4, 5]
    >>> my_array.remove(1)
    >>> my_array
    [2, 3, 4, 5]

Hey, this is kind of like a deeper hyperpolyglot.com! I appreciate the nod to learnxinyminutes.com, intentional or otherwise.

Could be pretty neat, but it won’t let me proceed to login with Google.

I was curious how categories and languages were managed. How do you prevent the inevitable “How do I hack The Gibson with Visual Basic 3” if they’re all fully crowd sourced?

Edit: mobile typos

Wow that's pretty sweet, another commenter mentioned learnxinyminutes but hyperpolyglot looks awesome as well.

I guess for now while it's small I have to heavily moderate, which I have no problem with. If something is detracting from the platform I will remove it, but it's a fine line where I cross into S/O territory where they mark everything as off topic.

I'm unsure what's going wrong with your Google login unfortunately. A lot of others have managed to register with no issues.

It's possible your Google email is not verified?

I just submitted a code block, and I didn't get any feedback after my response. Is there a manual approval process? I like the interface, though it could benefit from organization better.

This is actually really cool - imagine that you could go from pseudo-code to working program using this library.

No approval process. I can see your example here https://howtoxwithy.com/how-to/0b04d75d-e929-4daf-9c0a-57bfa..., just click the arrow button on the JavaScript panel.

Definitely agreed on organising better.

Quick request: Can you make the question a link?

“Oh perfect, how to append an array in Java is what I’m looking for. Tap. nothing. question rotated out already. Ok.. Do I have to type the whole thing in search to find it now?“

Hey, great idea! I will definitely do that

I can see that the project is very early in development but honestly this is great. I hope you use all the feedback here to improve it. You have a winner of an app, in my opinion!

Cheers! I really appreciate it Jesse

Sorry for the double post - but editing a snippet causes the snippet to disappear entirely. A bug tracker would be really helpful here :)

Definitely. I am tracking that here https://app.portabella.io/projects/daa40d3e-655f-4cb4-b4f4-6...

Thanks for the bug report!

Haha that website is also annoying. They're using the history API to manage the modals. So if you click that link directly, then click the X or simply click out of it, you end up back here.

EDIT: I literally can't even click on the site's logo to go to their front page without being kicked out of their site entirely.

EDIT2: Oh you made that site. Don't use the history API to go backwards (: it's a UX nightmare.

Yeah that's my other project, but good to know. I will sort that out.

Didn't mean to unleash a shotgun of criticisms on multiple of your projects, my apologies :D way more effort put into these than I normal put into side projects :) I would actually like to try portabella, hence why I was trying to click the logo :D

No worries, I don't take it personally.

Like I said in another comment, Portabella is my main side project. I whipped this up quick because I thought it was a cool idea.

I'll be launching Portabella on Product Hunt and Hacker News next week, this was kind of a trial run for that.

Any questions or issues shoot me at alex@portabella.io

I love how in the very first example, how to open a file in python, the example forgot to close the file!

Cherry on top. It's not as if python had a special syntax to use a resource and close it automatically at the end of the block, precisely to avoid this sort of issues. /s

Yeah it should be using a context manager.

403?

Seems up to me?

8 hours later, still:

403 ERROR The request could not be satisfied. Bad request. We can't connect to the server for this app or website at this time. There might be too much traffic or a configuration error. Try again later, or contact the app or website owner. If you provide content to customers through CloudFront, you can find steps to troubleshoot and help prevent this error by reviewing the CloudFront documentation.

Generated by cloudfront (CloudFront) Request ID: CYfM2liDaLwK2Mu8q7vnSE5rn89Fg8HmUSY6jIrL_5iTxe2dU1ADyw==

How will this be monetized?

Why do you presume it will be monetized?

I guess because earning money is nice.

This project serves a few purposes:

- scratch my own itch. I am missing something like this

- test launch on Product Hunt/Hacker News. I developed this in 6 days and I have a larger product I want to launch soon (https://portabella.io) but I want that to go perfectly. This served as a kind of dry run

I beefed up the servers for today but day to day I only expect this to cost me a few bucks a month, so keeping it running is no issue.

It can also be nice to point to something and say, I made that, so I'm happy to have it up and running.

I assume by your question that you presumed it to be run indefinitely at a loss? If that’s the case may I ask why you presumed that it wouldn’t be monetized? If that’s not the case and you’re asking a scholarly question for the sake of knowledge my answer is “because it’s not otherwise sustainable”

I've built close to 30 free small sites that are all running at a loss. Costs me about $100 a month in total. It's fun, rewarding and worth every penny.

Reply on news.ycombinator.com