Hacker News Reader

Comment by simon04 | Hacker News Reader

I'm one of the few maintainers.

Updating docs to a new release is easy unless the documentation system (such as react.dev redesign) or design is rewritten. Some projects seem to do this on a regular basis.

Some documentation generators generate random class names (such as .gtWOdv, .ezMiXD, .gOhcvK on docs.npmjs.com by Gatsby) which makes cleaning the docs from superfluous content (such as on-page navigation) very cumbersome and flaky.

Monthly, we auto-generate a list of outdated docs, here is the latest: https://github.com/freeCodeCamp/devdocs/issues/2105

Help is always welcome. :-)

View on HN

Hey simon04, I wanted to let you know that many many years ago the work of you maintainers made all the difference in my career and later my life.

Being able to read docs offline while commuting for some software I was pressed to work ended up being very important.

I just really wish you to know that albeit you may have not made a single $ by helping devdocs you are helping real human beings.

This is a very frustrating app for me. It is one of the best document sources out there but has become unusable because it cannot retain my selection of documentation. Almost every other time I visit I have to start from scratch picking the stack I use. It's great but not great enough to keep doing that over and over and over ...

I don't have an issue with dropping cookies or local storage elsewhere. I'm on an updated linux chrome. Any ideas?

Use devdocs-desktop? There's a lot of webview wrappers that have a separate config folder, and store docs offline.

https://github.com/hardpixel/devdocs-desktop

The enabled docs are stored in local storage. Are you frequently deleting your browser data? The enabled docs may also be exported as well as re-imported as JSON.

Could you rate the documentation generators based on how easy they are to consume?

I'd like to know how Sphinx, Docsy, MkDocs, Docbook, etc. compare in terms of being easy to semantically extract.

I'd need to look up the individual scrapers for a fair comparison, since I tend to forget/mixup the challenges. Some scrapers have been around for 10 years and only required little updates.

In general the more native HTML elements and the more descriptive CSS classes are used the easier it gets. Disadvantageous is when great parts of a doc page are built using JavaScript, e.g. when the whole nav is generated dynamically as the nav is typically the source for categorization/grouping on devdocs.

I was once asked in a technical interview how I’d do XYZ in some framework.

I said, “well, I’m not exactly sure, but I’d look up their API interface on devdocs.io to try and understand more…”

The interviewer had no idea what I meant, so we pulled it up on their laptop and they were blown away.

Granted… I didn’t get the job. But it was still pretty cool to spread some knowledge to the other side of the interview table for once!

Your contributions to this site is keeping it alive, and as a result it has inspired me to give a talk about my favourite updates to Python, since release 3.8. I could have found the data myself for sure, but you make it super convenient to compare all versions.

Well done!

Is there any discussion around eventually providing commenting like the PHP docs of old (I don't know what the PHP docs now look like off-hand as I write this)?

> I don't know what the PHP docs now look like off-hand as I write this

The styling and layout changed somewhat, but the content is pretty much the same, comments and all.

Today: https://www.php.net/manual/en/function.fgetcsv.php

15 years ago: https://web.archive.org/web/20081218125142/https://www.php.n...

Not that I'm aware of. In a way this contradicts the minimalistic and offline approach.

Are the docs sourced from HTML? Is it possible to ingest from the original source content e.g. .tex , .md etc?

How easy/tricky would it be to bring the pytorch documentation up to speed?

Thanks! I've literally had this tab pinned for years. It is awesome.

How can I help get playwright in there?

Here's how to add a new scraper: https://github.com/freeCodeCamp/devdocs/blob/main/.github/CO...

Or open an issue and wait for somebody else to implement the scraper.

Here you go: https://devdocs.io/playwright/

I'm thinking ChatGPT can probably do a lot of the heavy lifting in this cleaning step, have you considered integrating your app with OpenAI's API, or perhaps an open model like CodeLlama?

Do you realize how much money that would cost? "just throw AI at it" is often some of the most expensive words uttered lately.

"...but can't you just run it in the cloud?" :)

Would you run in your basement?

Probably a datacenter...

Your own?

Usually, you rent a cage/room in one. For all practical purposes, it is yours. I went into a few of our data centers at my old job. There'd often be entire areas dedicated to spares, scooters, or whatever people visiting there might need.

If you can afford to run in the cloud, you can often afford 2-3 times your max capacity in a data center. People ended up in the cloud because it was less expensive. Now that the demand for data center space has decreased, running bare metal is less expensive by a large margin.

Combine that with Kubernetes and Harvester, you can basically have your own cloud at a fraction of the cost. Now you just need actual sys-admins instead of AWS experts ... and they cost about the same.

So, in your opinion, bare metal is cheaper just because there's a demand/supply imbalance in the cloud business?

Why's there so much demand for cloud and, despite empty DC spaces you pointed, there's no demand to fill it?

And fixing all the tiny little mistakes that will litter everything

Isn't it better than fixing all mistakes?

It's probably better to just not have any mistakes to fix.

Isn't that wishful thinking?

Do you realize this app is used by OpenAI's target audience and they could be interested in sponsoring?

are they though? Seems like they're more interested in money than exposure at the moment.

Reply on news.ycombinator.com