GitLab Act 2 - A letter to our customers and our investors.
(about.gitlab.com)
from tofu@lemmy.nocturnal.garden to selfhosted@lemmy.world on 12 May 00:31
https://lemmy.nocturnal.garden/post/683024
from tofu@lemmy.nocturnal.garden to selfhosted@lemmy.world on 12 May 00:31
https://lemmy.nocturnal.garden/post/683024
Lots of layoffs (“re-evaluating our operational footprint”) and switching to “agentic” processes. Target user is AI.
Anyone still hosting Gitlab?
threaded - newest
They once were a promising alternative to MS GitHub but now they’re going down the same route.
Besides their pipelines being miles better, they never wete that great of an alternative.
2018 when Microsroft acquired GitHub it was the largest competitor with similar functionality as far as I know. I used if for a few years before switching to selfhosted Forgejo because selfhosted GitLab back then was painfully annoying and complicated to setup and used enormous amounts of resources and also felt more aimed towards corporate users.
Oh god that is so cringe. Just getting into coding i have no idea what to use as an online repo. I dont want to use github because microsoft but i want the basic repo collaboration features to be available cloning, pull requests, issues etc.
Host a Gitea?See reply comment below
Or Forgejo, which is a fork of Gitea and is what Codeberg uses. They explain their advantages over Gitea here: forgejo.org/compare-to-gitea/
The tl;dr is that Forgejo is maintained by a non-profit whereas Gitea is maintained by a for-profit company, and Forgejo is completely open-source whereas Gitea is open-core with some features only available in their hosted service. Forgejo also has better testing and a bigger focus on security.
Oh dang I didn’t realize! Thank you!! I was just starting to look at those things myself and wanted to also avoid GH. Plus Gitea was available on Yunohost too. I’ve heard of Codeberg, I’ll see if I can host that instead. It’s too bad other companies don’t move away from GH…
If you don’t want to host something yourself, check codeberg
I like codeberg and have no plans on migrating away from it, but their codeberg Pages product is…weak to say the least. There’s very frequent downtime. I had multiple users reach out to me letting me know my site was down… embarrassing. I set up kuma uptime checks on it, and now I see when the outages happen.
Forget “four 9’s” or anything close to that….my 30 day uptime is a measley 91%…
They communicate that openly tho:
status.codeberg.org/status/codeberg
Not sure if that’s also for the Pages feature, but in general having a weekly 8hr maintenance window is not optimal.
That’s true, but even at worst case (full 8 hour outage per week) that’s still 96% uptime.
Most of my outages have been out of that window.
Github seems to be down a lot, too, although perhaps not their pages part. Perhaps you could try to have just the pages in some other place?
That’s true! I’ve read recently that GitHub’s uptime is pretty terrible too.
My site is low enough stakes that I can live with it on codeberg. I just relaxed the uptime check a bit so it only alarms if I have an extended outage. Even so, the alarms aren’t actionable to me…other than maybe announcing to my users that there’s an outage rather than having them ask me if I’m aware of the outage.
Thats rough. Check out https://grebedoc.dev/ I believe codeberg itself also wants to migrate to that as well, I cant tell you how reliable it is though (I am using hetzner managed for 1.90€/m) is but I dig its simplicity.
Codeberg, I can recommend
Isnt codeberg centralized? I worry it will run into the same issue as github. I was checking out Radicle but its cryptic and hard to search for other projects.
Oh sorry, I might have misunderstood your question. Yes, Codeberg is centralised, but it is registered at a public e.V. in Germany making it more open (not a company).
But then you could use what they use, Forgejo to self host.
Or Gittea as suggested by somebody else.
It is but they’re working on federation for forgejo (which powers Codeberg).
Codeberg is supporting forgejo which Codeberg is built on. Forgejo is ActivityPub powered git repositories. So imagine regular git, but everyone can have their own repos on their own sites and you can still interact with each other. So yes, Codeberg is centealized FOR NOW. But they’re working on opening it up to EVERYONE to run their own and be able to access all the repos you use over the Fediverse.
Will it be possible to have decentralized pull request? Like I open a PR on my site, my friend reviews my PR on his site, and I get his reviews on my site?
That’s the plan, but it’s still far away
That’s nice
This was always baked into basic git from the beginning if you review your code in E-Mail chains or mailing lists.
So not really baked in at all then?
why wouldn’t it be? you can send emails from web uis too. you can share diffs however you desire. you can have a remote for each developer, and push/pull changes to each other. the github mindset kind of ruined the resilience and distributedness of git: one central remote, one account authority, one central place where discussing MRs… ever forgejo is not as good as decentralized git: what’s a forgejo identity?
meanwhile git has been decentralized and distributed since day one, linux is still developed in a decentralized and distributed way and forgepub is just not ready and not even close.
sending emails with an attached diff to many ppl is too hard? make a nice offline gui doing that and we’re distributed. github was a psyop to make us un-learn git, making it better is silly, like wasting decades searching for “good cigarettes”
Okay, but by definition none of that is “baked into” git…
E-Mail workflow is baked in
git send-emailcan directly send an email and every committer is identified by a mail address.Oh, well there you go. Looks like email PRs are baked into git.
That’s nowhere near as convenient as current web based PR.
Email chains and mailing lists are not really a practical way to develop anymore, and it is increasingly anachronistic (as is the idea of tying your identity to an email which is also baked into basic git). This was the only realistic democratic and federated option when git was designed, but it was never the ideal one. Forgejo is trying to build a better, more ideal, also-federated alternative that is really designed for code collaboration from the ground up. Once the design is stabilized, there’s no reason it couldn’t get built into git also. I would love to be able to create a PR with git itself and have it automatically submitted to the origin repository.
By what objective measure? Just because everyone listened to Github’s siren song, doesn’t make it so.
So many Open Source projects are just one or two maintainers. So many PR’s on Github involve just two people. That would all work over E-Mail, which is by design decentralized, unlike the web, which is by design not.
That sounds like the dream.
Just like bluesky is centralised “for now” i.e. forever
Except bluesky is funded by VC and they created their own protocol and federation design.
Codeberg is an open source repo only place, they’re building in AP, they have monthly updates. So nothing like Bluesky.
But I understand the trepidation.
Codeberg is a nonprofit, that has a democratic “member” system and is fully funded through donations / member fees.
The protocol extension is ForgeFed and it’s still a work in progress afaik
The issue tracker is on Codeberg.
Forgejo is only one of the implementations and not the reference implementation.
It will also be more general:
It’s funny coming from the Plex thread into this; ~100% of people who keep using Plex do so because it’s centralised and it makes sharing their library with their network of family and friends easier.
The truth is; a lot of us feel like we need more internet accounts about as much as we need genital warts. Part of the reason GitHub got successful was the fact that you only needed to register once and you had access to fork and PR all the repos on there.
Decentralisation is great for self hosting things for, well, yourself and your household, but it’s got hefty downsides. Account creation is a friction point for others to join and collab.
You are confusing decentralized and fragmented (or self hosted). The promise of fragmented software (like Lemmy) is that there are many instances but an agreed upon protocol. You create one account on one site and then use it to pull and push data to any other site that uses the same communication protocol. Like you and I for example. You created an account on lemmy.zip, I created one on lemmy.world, and we are both discussing a post created by a user on lemmy.nocturnal.garden (an instance I have never heard of).
The problem is, I have an account on lemmy.world but switched off during a time it had major problems with downtime and broken images. When I wanted to switch to another provider, my account was not portable. I hadn’t posted or commented an overwhelming amount, but it’s still not associated with this account.
So let’s say someone creates a federated Git hosting platform and feature matches GitHub with Actions/CI etc, so there’s no reason not to switch. Let’s then say git.world starts acting up, but you can create an account on git.zip instead.
Now you have given up your commit history and any commits you make from your git.zip account is not neatly linked with your git.world account.
I’m sure this problem can be solved, but it’s vastly more important for it to be solved before federated Git hosting can replace the “security” of GitHub. We do have to consider the fact that some people point to their GitHub profile when job searching, so git contributions and commit history is more valuable than Lemmy posts.
At least with federation a single account gets you access to all the systems. So a truly federated git system would be great.
Even if, switching your used repo hosting service is a matter of minutes if you’re using git. You register on the other site, add your SSH key, update the remote URL of your repository which is just a
git remote set-url origin <new url>and then hitgit push, probably with something like--forceor another option, kinda forgot the exact name. So that’s something you could easily automate in like 10 lines of bash script for all your repositories.It’s super hard to “trap” people in something like github because git is so open and decentralized. Switching is super easy. Most people who stay on github or gitlab do it because they need the CI/CD pipelines or because they’re lazy and/or stupid.
When I read this discussion on HackerNews they act like they’re trapped and it would require moving the sun and the earth to switch over.
Yeah sounds like a big nothingburger to me. If you just use gitlab for private projects with basic pushing and pulling without any fancy gitlab features, switching is a matter of minutes.
Now, if you’ve built your entire company setup around gitlab and use everything they offer, yeah switching is gonna be a lot harder and will require more preparation. However, it’s not impossible in the slightest. Even a large corporation with hundreds of developers could make a switch within 2 weeks.
And the open issues, tasks and pull requests?
Right.
Those are all part of the forge, not git.
The 2 are very different things.
And what kind of service is gitlab, which we are discussing here, or github which was brought up in the comment, or codeberg?
They are forges.
I think the comment of migrating git, was more for smaller and maybe private projects. Not large collaborations. So only the git part, not the forge part.
docs.codeberg.org/advanced/migrating-repos/
Those aren’t git features, those are features provided by surrounding tooling, not git itself, so I didn’t really consider them. I also never used them in private projects.
However, issues you can migrate easiely. I’ve seen tools out there that copy the issue content from github and to somewhere new. The creator of that issue is then a bot user or something, but the issue is still there and can be worked on. On github, the bot will leave a message that this issue is now handled somewhere else and closes it. Done.
Pull requests are also simple, you just merge them all. I haven’t seen a lot of projects with hundreds of open pull requests that were lying there for weeks or months. Now yes, you will lose the comments and history of the pull request itself, but I don’t think that’s very important.
Tasks I don’t know. I’ve never used them and don’t even know what they do. If it’s just a glorified kanban board with plenty of cards that say “Do X”, you can just copy paste them to your new tool because there’s nothing technical about them.
Its centralized, but they (forgejo, the underlying software) are building on standards wherever possible so it should be easy enough to move things around. I also don’t really see them breaking bad anytime soon, at some point you have stop worrying and start to build shit.
Codeberg or sourcehut.
Gitlab was always cringe.
Even cringe as it always was, still a better product than GitHub. Microsoft drove it into the ground, but it was quite bad before as well, comparatively.
For a beginner, I’d probably stick to Github initially, just because there’s so many guides and tutorials on how to use it, and their free plan is still pretty generous.
A lot of the knowledge is transferable though. If you do want to try something else, Codeberg is pretty good for open-source.
To just learn about Git, you don’t even need a host like Github or Codeberg. You can have a Git repo just on your computer, and still get a bunch of the benefits of source control - a full history of everything, separate branches and worktrees so you can have multiple incomplete changes and switch between them, etc.
Codeberg
Codeberg for hosted, Forgejo for selfhosted.
They are great.
This reads almost like a parody.
The only large mainstream competitor, which would probably benefit from github’s troubles: “We saw github breaking itself regularly because of it’s own slop coding AND flooded with trash vibe coded projects and thought - that’s where we wanna be!”
Gitlab and gitlab-ci really are great and easy to support with little problems as long as you update regularly. It really does look cringe, but they always were chasing current dumb thing relentlesly.
Yeah I like them as well, it’s what we use at work. The article doesn’t leave me optimistic though
Shameful behaviour.
An almost inevitable result of venture capital, IMHO.
All I see is layoffs and creating office space to force people to go to office. Well RIP Gitlab.
fucking drones
Sure bud
Acronyms, initialisms, abbreviations, contractions, and other phrases which expand to something larger, that I’ve seen in this thread:
6 acronyms in this thread; the most compressed thread commented on today has 6 acronyms.
[Thread #285 for this comm, first seen 12th May 2026, 08:30] [FAQ] [Full list] [Contact] [Source code]
objectively insane.
I still believe that’s not possible, but that’s only my opinion.
This part actually makes sense. Plenty of software doesn’t get written because it’s just easier or cheaper to do without it. It’s why BPM tools exist. Simplify the coding process and you can solve problems more cheaply.
I also think this will kill BPM tools. Why use BPM tooling when creating a real app is just as easy and more customizable?
I see that requires some more explaining my thinking:
There is only demand and supply.
Previously, we had “high demand” and “limited supply” which is what lead to software dev roles being a very well paid job in silicon valley and some other places.
Now, the promise of AI, making software by itself or increasing productivity, if true, mean that supply increases. That makes software cheaper. Theoretically.
But that’s the supply side.
What you’re talking about is also a “I have so much supply, I can now afford to do projects and software I could not do before, because my time, budget, etc. was limited.” But you already had the idea and the “demand” however low priority, already existed.
What isn’t happening, is that some company sits down and suddenly decides that they need more software than they thought they needed. Even the bit that is “replacing real humans” is replacing humans. It’s meeting a demand that was already there in a new way.
Using a metaphor / example, we currently, as humanity, manage to feed ourselves. Or let’s pretend that we do and nobody is starving. Someone claiming that “the demand for food is going to go up” is talking nonsense. They can say that demand for “cheese” or “meat” or “potatoes” will go up. But not food, because that market is already saturated. Because we’re not starving.
Yes, the fact that the demand is there and that the supply gets cheaper will mean that more software will be produced.
But not because of increased demand. AI doesn’t create it’s own demand.
…at least that’s my thought process and why I wrote what I wrote in the original comment.
I see what you’re saying. That makes sense. It’s an overloaded term.
In economic terms, a price drop would result in an “increase in demand”. Because current demand is the amount request “at current price”.
And that’s why it’s always talked about in relation to demand curves. Or how much demand there is at many theoretical.prices.
I don’t think that they’ve used enough buzzwords.
Increased buzzword utilization is part of their go forward strategy that begins implementation in Q3 pending socialization of relevant KPI. obviously.
Synergizing good EQ through the single pane of glass.
I’ll just leave this here.
This better?
This is so weird. They gave a Duo presentation at our company and I was a bit second hand embarrassed because it’s just bad.
Anyway, the stock price will probably go up after this announcement…
Stock went down though
Damn
<img alt="Woman clapping with the caption “nice”" src="https://leminal.space/pictrs/image/8d4940eb-fd2b-4d66-bfe8-94531c7039c7.gif">
My work started a trial of Duo and it actually was rather helpful. I liked it better than other teams just copy pasting output from gitlab to claude or something. Integrated well and had the option to hide all of it if you didn’t want any of it.
Interesting. This copy/paste approach was pretty much what they suggested we use. Either that or letting the agent do it’s thing completely on its own.
We use GitHub copilot now. And compared to that, Duo is lacking lot.
I would have liked it just for the integration with code reviews, but we didn’t even get a chance to try because “it’s too expensive” …… as we’re migrating from on-prem infrastructure licensed per site to cloud infrastructure charged per usage, and with no valid cost comparison
Ok what are we going to call them now? Gitslop? Sloplab?
<img alt="An excited energetic man pours water onto a steak, caption: “Let’s Slop 'em Up”" src="https://leminal.space/pictrs/image/1c5bd51b-e9e1-47c0-99a5-06f0653a6020.gif">
Forget they exist and don’t call them anything.
I like AI and use it. This post was just sad. What a crazy way to announce you don’t have an AI product while saying your product is dead.
Ouch. My company was just about to start moving over to GitLab off of Atlassian.
That makes sense, since Gitlab seems to be trying to challenge Atlassian. In who manages to make worse software…
We use atlassian at my job and I hate it
I like that I can read this as you stating you use Atlassian yet hate Gitlab, and the statement still works either way 😅
I always avoided gitlab as much as possible tbh because of how awful the interface is
also I just want to add that we just moved to on cloud atlassian and it is even worse
My condolences.
<img alt="friendly hug of two anime girls" src="https://leminal.space/pictrs/image/71d91f99-ba44-47e7-9e97-2718dff13430.gif">
After having to use their cloud version for a while I can confidently say that everything is even worse now 👍
<img alt="an anime girl that is looking nervous" src="https://leminal.space/pictrs/image/62c02d2a-4515-4ea6-aa6c-9f381bba21ad.gif">
Is every mother fucker just going bat shit insane this year? Goddamn it.
Codeberg has always been perfectly sane
The only upside I see is their stock has fallen since this announcement. Perhaps the market is finally getting that companies pushing AI isn’t a universal good thing?
Just like everyone has already done I’d also recommend Codeberg/Forgejo👌
Still hosting gitlab.
The CI on forgejo is, unfortunately, nowhere near as good.
Given how long gitlab has been struggling to fix basic bugs and instead creeping into features - hello-oo bloated and slow vscode-like web editor and non-ephemeral runner management - I’m not sure they have any staff left to let go. But it’s nice they found an excuse to shed their remaining talent and avoid complete stock devaluation.
“We don’t understand how the Dead Sea Effect works, and we want to super-size the damage.”. Okay, Bill.
FWIW Forgejo with Woodpecker CI is 👌 - a lot more like gitlab ci without the bullshit. It’s still basic by comparison though. I don’t think it supports templates
For fucks sake… Is Sourcehut ok for private projects?
I have been using sourcehut for mercurial private projects for about half a year without any issues. Also have some a couple of public repos which I develop in Mercurial and then mirror to Codeberg. Only issue I find with sourcehut is that they don’t produce files for users to download. So, if someone wanted something from your repo and they don’t have git / mercurial they would be unable to get the files.
I’ve also been using sourcehut for a few things for a number of years. Haven’t had any trouble. You do need to pay a small amount per month for an account, but I don’t mind throwing a few bucks to support an indie project like this.
Idiotic. Use codeberg.
MacOS CI jobs?
I always love to see companies do this with a semi open source product with investors
The code gets closed, a small clump of users split off, make their own version with beet and hookers, and soon the vast majority of the users following because the real open source one is so awesome
That was jellyfin’s story, but this is a variation on that and I’ve seen this story many times now
Bye bye gitlab,rest in pieces
I work on Jellyfin, but don’t like beets. Do I need to fork again?
Gitlab CEO - 16 years in Microsoft, Gitlab CTO - 13 years in Microsoft
Can we say Microsoft Gitlab ?
I hope not: we’re migrating from Gitlab to GitHub. I was never a fan because of the lack of enterprise features in GitHub (folders, with more granularity of settings and permissions, scalable usability), and certainly GitLab CI was extremely limited but wtf
Gitlab isn’t owned or controlled by Microsoft
No one claimed it was.
“Software will be built by machines, directed by people.”
Oh my lord. Is this a delayed April Fools post?
This is dangerous for me to say on lemmy, but fuck it.
Doesn’t it make sense that machines would write for machines? Isn’t that kind of what we already do by creating compilation layers programmers use? We obviously wouldn’t write the manual 1’s and 0’s, and most people don’t write using assembly. Is this not a translation layer for us to be able to write code?
Right now we have LLMs writing with languages designed for humans, and it’s already doing some pretty wild stuff. If we get to the point where AI is literally a coding model (and not a generic LLM) that is able to use an AI optimized way of writing code, who knows what it would be capable of.
Code is one of the few things AI is specially suited for. AI is just a big fancy prediction machine, so what better application than something that is by definition formulaic and patternistic like code? I am not saying we are there now, but rather the idea that machines should write software does make sense when it becomes actually feasible.
If we could have programmed like this from the beginning, we would have. There has been many evolutions of making it easier to code. What’s easier than plain language?
@Bazoogle @1hitsong First of all - when it comes to creating programs, you want the output to be deterministic. Stochastic program output is a serious problem, as you _will_ get unreproducible bugs. Second, plain language is _not_ easy except for the simplest of tasks. Actual programs need to handle all kinds of corner cases and hardware weirdness and human weirdness. Your "plain language" goes from "do a thing" very quickly to "do a thing. but not that thing. or that other thing. and and and.."
Your options would be write all those things in plain language, or program them all eith (hopefully) no mistakes, bugs, or vulnerabilities. Either way you have to catch all the situations. Even in plain language, not everybody will be able to effectively use AI to generate code. You need to have a solid understanding of software architecture to be able to get useful output.
AI is capable of writing deterministic programs.
I would also like to preemptively emphasize that AI is not there yet. I am simply talking about the concept of machines creating software. If you try to step back from your anti-AI gut reaction and truly think about it, it would make sense to do if we get there technologically
I don’t know why anyone should take your post seriously when you say that AI isn’t there yet. You’re saying, purely hypothetically, that AI could do these things, if it existed, which it doesn’t. That can’t be argued against because no matter what anyone points to, you can just say that isn’t it.
But, like, your basic premise that machines would be the best programmers of machines is inherently flawed because humans created those machines, and thus it should actually stand that humans would thus be the best programmers of those machines. But that’s a reductive argument that kinda is more tell than show.
Programming is really just some layer of abstraction on modifying how a computer works, so vibecoding should really be just another layer to that abstraction. But as it stands now (and how we have specifically created our current LLMs), these outputs are not deterministic, and thus sort of fail as a means to program with. That’s one of dozens of reasons of why it fails as a programming substitute.
I may be considering AI usage from a different angle. I’m less interested in the technical side than I am from the moral side.
AI companies trained their agents using open source software, did not contribute back to the code, did not credit the authors, and now want to sell it back to the same people they ripped off.
As an open source project maintainer, I’m disgusted by this.
I’m also a musician. AI companies trained agents on other people’s music without giving anything back to them. This also is disgusting.
AI trained on people’s work now lets you circumvent paying the original creators.
Add to this resource usage and environmental impact.
This is why I see AI usage as immoral. It hurts real people.
lol I just created my gitlab account today to get away from github and after reading this the account has been scheduled for deletion and now I have a new account with Codeberg. When are these dipshits going to learn that we don’t want AI in our workflows? I am capable of breaking things on my own, but at least when I break things I will learn from it.
Selfhost
Yeah I could do that, I have never done it before though which is why I was looking for another remote to use. Are there any good guides/tools out there you would recommend?
docs.gitlab.com/topics/offline/quick_start_guide howtohosting.guide/best-vps-hosting-germany/
Thank you!
Check out forgejo self hosting gitlab is overkill.
I just selfhost bare git repos on an ssh server
That’s a perfectly viable approach for users who aren’t interested in forge tooling and just need git push/pull capability.
Ugh, but I don’t want to buy a Mac mini for those CI builds
Can you clarify something about self-hosting for me please. If I want to collaborate with other people on self-hosted a git repo, do I need to have a static IP or a router device that is compatible with services like noip, and open up ports which opens up my local network to constant attacks from outside it?
It is usually easier to rent a vps with a domain to run as a reverse proxy or a tunneling server between your server and users. Dynamic DNS and static IPs depend on the internet provider offering real outbound access and not a NAT. Plus any internet facing service should be hardened for security in some way, which it is usually provided in service packages for the vps.
Run gitea on a small box.
.
Sucks, I manage GitLab in our company and it’s been difficult to maintain already without the vibe coded shit updates that break everything. I’ll need to see what are our options our but my assumption is that there aren’t any.
Forgeo?
The enterprise features we rely on do not exist in Forgejo last time I checked (8 months ago). Maybe it improved. Hell, my company would probably even be onboard moving to Forgejo if we can get a support contract with them and some of the enterprise features we rely upon (SCIM being the main one).
Here’s some offers on professional support: codeberg.org/forgejo/…/issues
<img alt="" src="https://lemmy.world/pictrs/image/404a08f2-cdbd-403e-9ee6-0a7ea1ad9054.gif">
They endlessly tooted their horn about their diversity and fully remote operations. So this is pretty rich.
“This isn’t cost cutting” Oh, fuck off. This is trimming the fat before they try to look for a buyer again.
<img alt="" src="https://lemmy.world/pictrs/image/d17f094c-c73e-475c-8163-e061b38b9e38.gif">
Translation: please buy us, we want to cash out before the bubble fully pops.
Forgejo is great.
Hmm I transferred a lot of my stuff from GitHub to GitLab a while ago mainly because of the free private repos. Was thinking about moving fully off GitHub but glad I waited 🙃 Still going to do it but gotta figure out which providers actually have their heads somewhere other than up their ass