2

GitLab plans to delete dormant projects in free accounts (Register)

 2 years ago
source link: https://lwn.net/Articles/903858/
Go to the source link to view the article. You can view the picture content, updated content and better typesetting reading experience. If the link is broken, please click the button below to view the snapshot at that time.
neoserver,ios ssh client

GitLab plans to delete dormant projects in free accounts (Register)

[Posted August 4, 2022 by corbet]
The Register reports that GitLab is planning to start deleting repositories belonging to free accounts if they have been inactive for at least a year.
GitLab is aware of the potential for angry opposition to the plan, and will therefore give users weeks or months of warning before deleting their work. A single comment, commit, or new issue posted to a project during a 12-month period will be sufficient to keep the project alive.

(Log in to post comments)

GitLab plans to delete dormant projects in free accounts (Register)

Posted Aug 4, 2022 21:20 UTC (Thu) by joey (subscriber, #328) [Link]

This made me curious how many projects have a 1 year gap in commits. Here's a one-liner to find them:

git log --pretty='%at %H' | sort -rn | perl -lne 'BEGIN {$o=time()}; @x=split " "; if ($x[0] < $o-31536000) {print $x[1]}; $o=$x[0]'

One of my projects, pdmenu, has 5 such gaps in its history. I still hear from users of it.

GitLab plans to delete dormant projects in free accounts (Register)

Posted Aug 4, 2022 21:34 UTC (Thu) by joey (subscriber, #328) [Link]

Hilariously, this includes some of gitlab's own dependencies, like the ruby seed-fu library which has not been updated since 2018.

No need for Gitlab to worry though ... that's hosted on Github. ;-)

GitLab plans to delete dormant projects in free accounts (Register)

Posted Aug 4, 2022 23:48 UTC (Thu) by dskoll (subscriber, #1630) [Link]

One of my projects, Remind, which is currently pretty active, had an almost three-year gap from 2001 through 2004. Earliest commit is from 1996 which obviously predates git or svn... but I still actively work on the software.

Anyway, I self-host my projects, with mirrors on salsa.debian.org and codeberg.org. Debian, in particular, is an organization I trust to act in the interest of software users and developers.

GitLab plans to delete dormant projects in free accounts (Register)

Posted Aug 4, 2022 21:51 UTC (Thu) by tchernobog (subscriber, #73595) [Link]

I am also a bit curious about those projects that are linked to the account of a deceased or otherwise incapacitated developer.

Gitlab is still relatively young so this issue is not that felt. But wouldn't it set a risky precedent?

It would increase the chance that long-term archival of code which remains interesting for historical reasons (be it legal, educational, or otherwise) gets dropped inside a massive black hole.

Considering how much we still rely on old dependencies even in relatively young projects (or in older projects: look at the amount of COBOL code still running on public administration servers/mainframes; tomorrow it'll be Java), the prospect is a bit scary.

Imagine being a government who has acquired the source code for some tool they commissioned. Some years down the line, they require some change to be made. "Whoops"? Taxpayers' money wasted again.

Or medical devices firmware, who are suddenly vulnerable, and the vulnerability lies in a component for which the source code has been killed. I wouldn't want my peacemaker to be incapable to receive a security update in a reasonable time frame.

Of course, one could argue that as a responsible company, you should keep a mirror of all your dependencies anyway. Practice however suggests that a lot of companies fail to do so, especially when handing over their source code because they are compelled to release it to the final customer by some contractual clause (for instance, in the EU).

GitLab plans to delete dormant projects in free accounts (Register)

Posted Aug 4, 2022 22:17 UTC (Thu) by pabs (subscriber, #43278) [Link]

This is one reason I like using distros like Debian, there is at minimum a copy of all versions of all packages in both binary and source form on the Debian Wayback server. That misses upstream VCS/issues/discussion etc of course. For the VCS of dependencies, Software Heritage has an API people can use to save code repositories. For issues etc, you would have to use the API of the service. For GitHub, there is github-backup, but I haven't found anything for GitLab yet.

https://snapshot.debian.org/
https://www.softwareheritage.org/2019/01/10/save_code_now/

GitLab plans to delete dormant projects in free accounts (Register)

Posted Aug 4, 2022 22:13 UTC (Thu) by pabs (subscriber, #43278) [Link]

I note that both Software Heritage and ArchiveTeam are working on archiving the affected projects before the deletion. Software Heritage already archives all of GitLab, but some projects aren't yet archived or haven't been re-archived recently.

GitLab plans to delete dormant projects in free accounts (Register)

Posted Aug 5, 2022 2:15 UTC (Fri) by IanKelling (subscriber, #89418) [Link]

GitLab Inc should do something like donate 50% of their cost savings to software heritage.

GitLab plans to delete dormant projects in free accounts (Register)

Posted Aug 4, 2022 22:14 UTC (Thu) by Smon (guest, #104795) [Link]

On one hand - only accounts with 1-year downtime.
On the other hand - I host my projects elsewhere. I don't want my source code be gone, only because I am gone. I hope my projects will outlive me.

GitLab plans to delete dormant projects in free accounts (Register)

Posted Aug 4, 2022 23:30 UTC (Thu) by dtlin (subscriber, #36537) [Link]

https://archiveprogram.github.com/ will be "forever" even if GitHub deletes those repositories now. Not sure how retrieval may work, though.

GitLab plans to delete dormant projects in free accounts (Register)

Posted Aug 5, 2022 1:15 UTC (Fri) by atai (subscriber, #10977) [Link]

Don't give Github more data to train, please

GitLab plans to delete dormant projects in free accounts (Register)

Posted Aug 4, 2022 23:51 UTC (Thu) by Depereo (subscriber, #104565) [Link]

It would be quite simple to create a scheduled gitlab CI/CD pipeline job that opens and closes an issue on a project once a month or once every six months, keeping the repository 'active' permanently using the platform's own tools.

Perhaps they will simply drop this plan, but if they do not it may become quite popular to use a base project template that includes 'keepalive' jobs.

Old man yells at cloud

Posted Aug 5, 2022 1:12 UTC (Fri) by smammy (subscriber, #120874) [Link]

I'm always curious about folks who seem surprised when service providers make their free tier less attractive in annoying ways. A free tier is a loss leader, and when the losses exceed the leads (so to speak), a company is always going to put the squeeze on.

That said, it's frustrating that they're pointing to hosting costs as a reason for doing this; are abandoned projects really that much of a traffic and storage burden?

Many mature build systems that pull down third-party source have a list of alternate mirrors to try for each package. The package managers for "modern" languages kinda act like everything's always going to exist in it's canonical location forever.

I think eventually we'll get used to authors making unfortunate hosting decisions, our tools will grow source hoarding, automatic mirroring, etc., and social expectations for authors will start to include making your code available in multiple places.

Old man yells at cloud

Posted Aug 5, 2022 2:31 UTC (Fri) by rjones (subscriber, #159862) [Link]

I don't know why people think that when people offer free services that they are then indebted for the rest of eternity. They don't owe you anything because you use their servers.

If you think that it's gitlab's job to host your content for you for decades without any input or payment it is pretty unfair. It is expensive to host all that content and even more expensive to share it. Yeah sure it's loss leader and all that, but nobody is going to host data if it means losing their business or going into poverty either.

It's really difficult for gitlab to know what are junk accounts people have abandoned and which ones are things that people still depend on. And if Gitlab goes through the trouble of figuring all that stuff out then it just increases the maintenance burden, the costs, and the likelihood that people will just take advantage of them.

If people want to host content indefinitely we are going to have to come up with a better scheme then entrusting it to corporate services that may or may not be around in 5 or 10 years. There is no guarantee that big corporations like Microsoft will be around in 10 years, much less lesser funded efforts like Gitlab.

Even entrusting Debian is no guarantee in the long term.

So not only you need a superior scheme, but you also need a way to fund it.

Luckily we have people working on projects like IPFS, IPNS, and related projects like Filecoin and the rest of them. This may solve the problem. Where you not only have a scheme to potentially host content indefinitely, but may have a actually useful way to pay for it by providing storage resources to the community. Essentially you provide storage resources for other people to use so you can then go out and use other people's resources for mirrors, backups, and other purposes.

It should be possible to do something like host Git repositories on a IPFS backend.

It is kinda possible now, but I haven't tried it. https://docs.ipfs.tech/how-to/host-git-style-repo/ At the very least it should be useful for tools like npm, go, pip, etc.

As long as somebody somewhere keeps a copy of it then it'll get preserved. You don't have to depend on people maintaining DNS records and web servers in such a way to preserve URL paths. Don't have to depend on infrastructure to setup and maintain mirrors of websites either. Just have to have people that are willing to avoid purging their caches after using stuff.

As long as somebody somewhere is finding the software useful it should be available to the rest of the world.

Old man yells at cloud

Posted Aug 5, 2022 3:31 UTC (Fri) by Vipketsh (guest, #134480) [Link]

> folks who seem surprised when service providers make their free tier less attractive

While I agree with you and understand where you are coming from, the other side of the coin is that free/open source software and communities can only thrive if there is free hosting available for individuals. Put another way: there is little chance some code will be made public if to do so requires it's author to pay the real cost of hosting it.

It would be great if there was "github for open source" or some such reputable non-profit around but reality is that general free hosting only seems to come about as a necessary evil for commercial entities to channel paying customers and that leaves the open source community with one option: extort the commercial entity with bad press/reputation if they do unwanted things with their free offerings. Not nice, but that's how it is.

GitLab plans to delete dormant projects in free accounts (Register)

Posted Aug 5, 2022 1:30 UTC (Fri) by pabs (subscriber, #43278) [Link]

GitLab have decided to keep dormant projects but put them in slower storage instead:

https://twitter.com/gitlab/status/1555325376687226883

GitLab plans to delete dormant projects in free accounts (Register)

Posted Aug 5, 2022 2:25 UTC (Fri) by IanKelling (subscriber, #89418) [Link]

This is fine, but I still don't like gitlab and I still like the idea of them donating some fraction of their savings to Software Heritage.

GitLab plans to delete dormant projects in free accounts (Register)

Posted Aug 5, 2022 2:32 UTC (Fri) by salimma (subscriber, #34460) [Link]

I guess I'm moving my CV away from gitlab then, ugh

About Joyk


Aggregate valuable and interesting links.
Joyk means Joy of geeK