I'm gonna complain here because I see gitlab employees commenting.
It's a pretty bad look to claim that you offer a free version for individuals when there's not actually any way to get it from gitlab.com. Your only option is an email capture for a free trial of enterprise. Even the install docs redirect you to an email capture.
Of course there is a way to install enterprise for free, and it's full of nags and enterprise-only features that prompt you to upgrade.
You have to already know what CE means and you have to find it through a search engine. Even the install docs barely mention it. I got all the way through the install process and configuration before I realized I'd installed the enterprise version. I only found CE by searching for ways to hide the enterprise upgrade nags.
It's an extremely bad look, and people like me who care about privacy and open source are gonna be turned away. I explicitly decided against gitlab for this reason. I only came back because the other options are worse.
I get wanting to promote the paid offering, but completely burying the actually-free version and only offering an enterprise trial behind an email capture is pretty hostile. It's against the spirit of FOSS and I'm very disappointed.
I understand your perspective on this, and I’m certainly not saying it’s wrong. But there’s some background you should consider.
GitLab CE instructions used to be the default. Paying customers would find they installed GitLab CE by default and run it, and then want to use features. They’d have to install GitLab EE over it. Well if they’re a year or two behind in upgrades, that transition can be painful and require services.
This created alot of anger and frustration from enterprises who paid for GitLab or wanted to convert to paying for GitLab. The solution was simple; install GitLab EE by default per the instructions. Because FOSS folks will search out the free editions. Yet customers won’t; and they’ll get caught with the CE edition unable to migrate to EE.
The move wasn’t one built on bad faith to move people to paid versions; or somehow bury the CE versions. It was to reduce paid customer frustrations. Even now GitLab docs talk about running GitLab directly from source.
They offer everything in Ultimate, including 50k hosted CI minutes, and the requirements seems reasonable (basically, everything needs to be public, have an open-source license, and you can't profit from add-ons or services).
The drawback is that it looks like you have to apply for it to use it.
Neat, I only wish their pricing for commercial was more reasonable. The bundling they do makes it a complete non-starter for people or companies who don't use the CI/CD, for example.
Another unfortunate thing of their Open Source Program is that companies that make open-source (MIT, GPL) and have a commercial offering as well (eg: GitLab themselves) are not allowed to use their Open Source Program.
I've followed the gitlab migration and every package and distribution change that warranted community notification for more than a decade.
It's such an empowering feeling to have tracked all the changes to the distribution over a decade. The Arch maintainer culture has managed to provide consistent high quality communication and documentation.
Most of the news doesn't require action on my part, being a subsystem or package I don't use. They use the news channel sparingly and the distribution is minimal and clean. News arrives only every other week or so and is succinctly written in one or two paragraphs.
It's a distribution for those who love precision and professionalism.
That's one thing (of many things) I've always appreciated about Arch: for those rare instances where I run `pacman -Syu` and it fails, I instinctively go to their homepage and look at the "Latest News", which invariably tells me what I need to do.
How many distros now have their own GitLab instance? I know Alpine, Arch, and Debian do. Any others?
Plus Freedesktop, GNOME, U-Boot, and probably more I am not aware of.
At this point, even if GitLab decided to close down their CE offering, I think there is enough momentum in the community to keep it going (at least for a little while).
Plugging in for GitHub gets a score of about 90%. GitLab is in the mid 30%'s. (Assuming Arch is keeping a backup mirror of the bugtracker, etc on a private instance somewhere.)
That seems good enough to not bother with actually hosting it yourself. I'd rather they spend volunteer time on stuff other than applying security patches to a giant beast of a web service.
A lot of effort is spent on packaging the stuff we need for our own infrastructure so we can run it in prod. Gitlab is one of these exceptions but generally it's been working well for us.
VLC is also using their selfhosted Gitlab. And just as on the Arch gitlab, they had to turn off autoregistration due to spam and are resorting to manual validation instead. Sad!
yeah but many FOSS orgs are run in a way that they state in their own bylaws that they would not use proprietary infrastructure and communication tools in order not to scare away Free Software purists. Then they get overrun by spammers and the whole project might get laid on ice because of it.
Honest question, I'm genuinely curious, why would you move from Arch to Mac OS?
I'm a big Arch fan but moved to Fedora because the overhead of managing updates in a rolling release got to be too much, and Fedora is very similar to Arch in conventions it follows, so most of my knowledge transferred over seamlessly (other than package manager of course). Moving to Mac OS though seems like it would render most of your knowledge/experience useless. I got stuck on a macbook for a short-term contract and found it maddening and confining in so many ways
The truth? Battery life. I was tired of 1-3 hours of battery life when I needed to take the laptop out, and a Macbook was the only thing at the time that gave me 7-15 hours minimum. That was a few years ago, and so the landscape has most likely changed since then, so I may look at getting a beefy machine and using Arch again in 2024.
Also, I guess it really doesn't matter in the end because all my development happens in Debian containers anyway. Hell... I could even use a Surface Pro and it would only slightly annoy me lol - the absolute beauty of containers!
Fedora? Maybe it was because of rpm prerequisite hell vs apt-get/pacman's auto install of whatever is needed, but moved away from RedHat after around 5.0/6... but then again, I'm sure that's all changed too. Maybe I'm just old :)
Congrats! Arch is my first and only distro and it's amazing.
Looking forward to possibly contribute.
I know this is off topic but does anyone know how are we on having packages with lto by default and multiple -march? Is the new pkgctl going to allow multiple automatic rebuilds?
So true! I love the AUR, it is like the Homebrew of Linux. Or, flipped, Homebrew is like the AUR of Mac.
My new company is considering adopting Linux and is considering Ubuntu, and I pitched Arch because I really want to use the AUR again. I don't like installing PPAs with Ubuntu.
I'm seriously impressed with Gitlab's issue system. I mean, it's apparent that they are great at git hosting and repository collaboration but the issue management system is really good. Epics, backlogs, stories, labels, tags, milestones, boards, roadmaps, it's all there.
I was just having a conversation with a coworker about Jira vs Gitlab and why it doesn't make sense for us to switch to Jira over using Gitlab issues. After test driving it for a few weeks and now with a little "nudge" of backing from Arch with this post about how they have migrated to it, I was able to convince my coworker that Gitlab issues is more than enough for what we do and that adding Atlassian on top of that would only add to costs. FWIW, I love Atlassian Jira+Confluence duo. I don't like Bitbucket at all. It feels very antiquated and Jenkins-like. The UI theme isn't the issue, it's the layout, the nav, the construction that just rubs me the "we don't care about your productivity" way.
Valve is a Linux and Arch MVP, and as a die-hard Linux user, my appreciation for them is immense. Without Valve's efforts, Linux user base would be a lot smaller because a lot of people would be using Windows. I wouldn't be able to game (since I am a Linux-only house) without a complicated windows VM setup. So just the fact that they support Linux with their proprietary software, and the fact that they used Arch for the Steam Deck (giving strong commercial incentive to game devs to support linux or at least proton, which IMHO is good enough) is a big contribution to the community. They also made the Steam Deck fully hackable by the owners, so you can do neat things with it including running all your games from some other store if you want.
But on top of that, they contribute a ton of things that benefit even people who never touch Steam. Valve slings graphics stack and GPU code, and of course Proton/Wine, Gamescope, and a handful of other things. They are awesome about upstreaming stuff, and when it's not upstreamable (say for example a new or standalone project) they tend to open it.
Proton, Gamescope, Linux ESync/FSync Patches are all at least related to valves efforts from what I can gather and as an Arch User I feel it's the premier linux gaming distro due to valves efforts.
I think most of the work Valve does is upstream from Arch though but it's trickling down very fast (with arch, sometimes hours, if it's an AUR-Git package, seconds).
PS: The core of Valve's technical contribution is around the SteamDeck. You already know it runs Linux, Proton, yadayadayada. An obvious fact that only clicked for me after being mentioned in the presentation: if you aren't in the Steam store but want your app on the SteamDeck the primary option is a Linux version (distributed through flatpack). That (commercial) incentive alone is a huge benefit. Suddenly you can make a business case why your product maybe should support Linux.
¨Apart from sanitation, the medicine, education, wine, public order, roads, the fresh water system, and public health ... what have the Romans ever done for us?"
> I feel it's the premier linux gaming distro due to valves efforts.
For anyone reading this: virtually every piece of code is upstreamed to the projects themselves (Mesa etc.), there is no reason to jump to Arch for gaming.
I'm less likely to trust "open source but kinda not" platforms like GitLab. It's only a matter of time before more options get locked down and you're looking at paying for a code forge.
Git is free software and always has been. All these forges being split between community and enterprise are just so clearly people trying to make money on the popularity of other technology.
There don't appear to be any one-time purchase options, either. So, a subscription so you can collaborate with other people... Yeah, no.
cgit and Mantis and bugzilla all still work and are actual libre software.
Not really. FSL has some gray area surrounding the expiry of non-compete, and I expect companies to extend their non-compete term whenever revenue gets low or a big feature gets released. Also, wouldn't this period be reset on every new version of the software, so there's a rolling 2 years you can't fork or copy?
Their handling of the switch to systemd was dismissive. Tom Gunderson simply left sysvinit to rot. There was no grace period of switching, and sysvinit was removed prematurely. It was one of the first distros that went all in on it, well ahead of others. They claimed that sysvinit would remain a choice but did not clarify how long that choice would remain and generally just made the switch and said "get over it".
I did, I left Arch Linux and went to Gentoo where I could have control.
Arch is a shadow of what it was under Judd. phrakture was the worst thing to happen to that distro's culture.
Thing is, sysvinit, runit, s6, and OpenRC all work too. It's not just systemd. The only reason we 'had' to go to systemd is systemd was actively being sold to distros and they were implementing things to get them to switch.
Personally I find that suspicious behavior in an allegedly open ecosystem.
Yes, the other big new thing also is that people can now create MR's for the packages aswell!
So far this is working well and we are getting a lot of high-quality contributions :)
Have the guides been updates yet? I created my first one but it wasn't obvious whether things like pkgrel should be increased as part of the MR (I assumed not) and in what format to submit them/which information to include in general.
Gitlab is a great product but using it has a negative effect on getting contributions for an open source project outside of the core developers due to Github's larger user base. I have seen this play out many times.
I don't buy into this argument. If someone doesn't have the knowledge and motivation necessary to use another platform, they aren't going to be making meaningful contributions to your project.
It might mean less bug reports from users who find bugs, but aren't deeply affected by them.
The upside is a lot less garbage in your issue tracker.
You're forgetting the users that are tired of signing up for one-shot sites to do patches or drive-by bug reports. The actual value of the sign-up process is so little that the incentive to contribute lowers. It's another set of username/password to manage. Even if you're unwise and depend on a password app, it's another thing on the list to manage.
OpenID was supposed to fix a lot of problems and doesn't appear to have, at all.
Anyway, there are lots of people who are not only knowledgeable enough but also skilled enough to triage bugs, but don't want to go through the rigamarole of a new account, e-mail, clicking a verification link, reading the bug report guidelines, etc.
Putting walls up to contribution is indeed a good way to get fewer people contributing. There's no guarantee the ones stubborn enough to still report bugs are reporting any better, though.
I'm glad it's not just me. I've given up on bug reporting to Debian on at least ~two occasions because the email sent via reportbug simply wouldn't go through without any indication as to why. Fortunately somebody else was more successful and all got fixed.
It's a pretty bad look to claim that you offer a free version for individuals when there's not actually any way to get it from gitlab.com. Your only option is an email capture for a free trial of enterprise. Even the install docs redirect you to an email capture.
Of course there is a way to install enterprise for free, and it's full of nags and enterprise-only features that prompt you to upgrade.
You have to already know what CE means and you have to find it through a search engine. Even the install docs barely mention it. I got all the way through the install process and configuration before I realized I'd installed the enterprise version. I only found CE by searching for ways to hide the enterprise upgrade nags.
It's an extremely bad look, and people like me who care about privacy and open source are gonna be turned away. I explicitly decided against gitlab for this reason. I only came back because the other options are worse.
I get wanting to promote the paid offering, but completely burying the actually-free version and only offering an enterprise trial behind an email capture is pretty hostile. It's against the spirit of FOSS and I'm very disappointed.