I use youtube-dl to download from Nico Nico Douga, which means having to restart the download every few minutes because the program doesn't send heartbeat requests. There's been a PR to fix this for a couple of years now [1], and a more recent updated PR to add encrypted video support [2], both of which are still open. Meanwhile, the corresponding PR on youtube-dlc [3] is actively being looked at.

[1] https://github.com/ytdl-org/youtube-dl/pull/18230

[2] https://github.com/ytdl-org/youtube-dl/pull/23824

[3] https://github.com/blackjack4494/youtube-dlc/pull/55

It makes sense that it's easier to give <100 PRs attention than 25k+. If the DLC project grows, you can't know if the same won't happen again.

Not saying forks should never happen - streamlink and KeepassXC are examples of great forks off the top of my mind - but you can't judge just on anecdotal stale PRs.

> It makes sense that it's easier to give <100 PRs attention than 25k+.

ydl has 700 open PRs, not 25k+. ydl hasn't had 25k+ PRs in its lifetime either, it's had 4k: GH issues and PRs are the same object, so the sequence is shared.

Doesn't change the point. All of the fork's 55 PRs except 2 and most of the issues were made by the forker himself as well.

It's not like the forker inherited all the open issues/PRs, they just dropped them all.

Getting popular and having the masses leave a bunch of issues/PRs is hard. Reviewing code is hard and takes more expertise than writing it. A lot harder than reviewing your own PRs on a motivated (for now) fork.

> It's not like the forker inherited all the open issues/PRs, they just dropped them all.

That's wrong, most of the recent commits in the fork are merging branches from a lot of other repos. GitHub doesn't have a mechanism for inheriting a project's pull requests when you fork, but that doesn't mean they're automatically "dropped", since you can still merge the corresponding branch.

https://github.com/blackjack4494/youtube-dlc/commits/master

Actually I merged quite some PRs that were open or even closed. However I tested and reviewed all I can but those geoblocked ones. But most of these were verified already or I could use some simple self hosted vpn. But do note that my reviews are not really in line with the strict coding conventions originating from the main project. As long as it works and isn't totally gibberish it will be approved by me. I rather want functionality and somewhat stability quite fast.

Although totally agree reviewing can be such a pain..

I maintain several open source projects. Sometimes PR are not merged for years.

I'm not payed for this, I'm not even thanked for this most of the time, so I'm certainly not going to set any deadline.

But forking might be a solution, if you can give the necessary horse power for the long run.

Most can't.

It’s odd to me how nontechnical people have migrated to GitHub to make bug reports on software they use.

The other day, I was trying to use a Python library to scrape some stuff. It kept breaking. I looked on the issues page and there were many, essentially useless bug reports asking them to fix the issue. I opened up the script and changed about 7 lines and it worked fine.

There are two factors at play here I think. The first is that Github has become quite popular. The second is that Github has worked a lot on usability - it's now very easy for users who are not very technical to create an account and report issues.

At times I imagine there could be some kind of "I know how to report bugs"-certification that people could somehow earn. Projects could then use this as a barrier to prevent getting too many useless issue reports.

The problem really goes both ways - as a software dev I'm disincentivized to report detailed bugs to projects, because there's a high risk it will get drowned out in the noise.

Some projects have quite specific templates for reporting issues. This helps a bit, but you still need someone to go in and remove issues that don't match the template, and few people want to volunteer for that work.

Yeah, I guess it’s just an entirely different kind of engagement.

I mostly use GitHub to find off the shelf solutions to whatever problem I’m having or to avoid doing the work myself. If I run into any problems I have with something from there, I either try and fix it by tweaking the repository, using another, or just building the thing myself. I think the last time I reported an issue on an external repository was because of something I was made to use at work.

Well, they look for a channel to communicate their frustration. Most are closed. This one is open. The water follows the path of least resistance.

I've noticed this trend too -- but I've also noticed people starting GitHub Repos for websites/apps just for the Issues.

No code, no PRs, just a way for the public to create Issues.

Not sure what this means for the "culture of code" other than things are blurring...

At least there is a feature called 'discussions' coming soon to github. That should hopefully minimize the abuse of issues. But that is also due to the fact of current github. Not everyone is displaying their email and there is no other communication channel.

I think a lot of it is maybe bleed over from gaming communities, especially emulators, maybe rom hacks too? I’ve noticed barrages of useless bug reports on those repositories. But I’ve been surprised to find them on far afield stuff as well.

There is an issue preventing the download of Watch Later playlists that has been open for over a year. The core maintainer, dstftw, closes all reports related to it with "duplicate" while never giving a reason or the original issue. See this issue I filed for an example.

https://github.com/ytdl-org/youtube-dl/issues/23860

He closed it without even addressing the issue of him closing the related issues that I explicitly brought up. There are scores of issues describing the same issue, but all of them are closed, and more of them are still being submitted. This has been going on since April of last year.

It's making me realize that sometimes a fork really is the only way to get around the problem that you're not in control of the project and there are bugs that many people need to be fixed.

Closing duplicated issues looks okay to me. The author not interested in fixing this issue himself looks okay to me. Is he refusing fix proposals about it?

Yes; there are now several hundred pending PRs for both fixes and new extractors, all ignored.

> Seems to have been originally spawned from this discussion: "Youtube-dl is dying?" https://github.com/ytdl-org/youtube-dl/issues/26462 citing many PRs are not getting merged.

As someone who has tried to contribute to YouTube-dl in the past, the absolute disinterest I received and complete lack of follow up on simple bug-fixes caused me to completely dismiss YouTube-dl as a FOSS project worthy of my time.

And I say that as someone who sends PRs to projects I’ve visited once, over tiny things like typos or lack of installation/build instructions in the readme.

Some of my PRs received a crazy all-or-nothing treatment, asking me to fix huge, long-standing bugs nobody had fixed for years, instead of accepting a 2-line diff which did a provider-specific workaround which solved a real bug, but seemingly not in the “right” way.

Completely toxic and hostile.

That someone has started a community-fork with a more welcoming and pragmatic approach does not surprise me in the least.

I had basically the same experience. I even got requests for functionality that was already there, as if the maintainers didn't even read or test the patch. There are a number of pull requests of mine that have been waiting open for years, ready to be merged.

Maybe I'll have more luck with this fork.

Counter-anecdote: Even YouTube support is somewhat flaky (age-restricted videos, private (as in, locked behind auth) videos, random failures), and several sites I use are broken and require additional options (cookies) for even basic functionality.

Which sounds perfectly normal for a webscrapper.

Are those issues you describe fixed in not merged PRs in youtube-dl and fixed in merged PR in youtube-dlc?

I find it gross that dstftw (the core youtubedl maintainer) censored the part about youtube-dlc from that issue.

CTRL+F'ing `youtube-dlc` comes up with several hits - was there more that is missing now?

You can look at the edit history of that comment. You'll see that dstftw edited it to remove references to the youtube_dlc repo.

https://github.com/ytdl-org/youtube-dl/issues/26462#issue-68...

Unfortunately, can't find a way to link to the edit in question directly. You'll have to click once there to see the diff.

Being able to edit people’s comments on GitHub is nice as a maintainer but removing information and completely changing people’s comments is really nasty. Any such effort should be clearly justified, preferably with a visual “edit message” that goes along with the smaller “edited” text.

I don't think that editing comments that you didn't make should ever be a feature. This is something that could easily be abused to make it appear that someone else said something they didn't actually way.

I can see what you're saying in principle, but what if a bug is opened with "this stopped working for me" but evolves into a very specific issue? Should the maintainer open a new bug, link to it and close the old one?

In practice, I find it really useful to tidy up the top message of a bug report (but /really/ like seeing history of it). I also see cases where users spam raw logs into updates and tidying it up (perhaps making the log portion an attachment or changing formatting) makes it much clearer for future-you and other collaborators.

Then there should be a "non-message" textfield up top, that doesn't pretend it's the original report, or forge its author.

Check the edit history of the first post.