ive been using gitlab for 3 years now in an enterprise environment, and the only complaint I have is lack of federation (albeit im told after the MS move, this is getting a lot more attention.)
I have 1400+ projects and more than 700 users in 3 geographic locations using my server. backups with ruby fog libraries included in gitlab are taken once a night and im expanding the CI presence to include a kubernetes cluster. In short, Gitlab is worth it if you have the resources to deploy.
I have also been using gitlab for 3 years. The main complaint I have is usability / UX. For example, if you go through their merge request flow, parts of the page are unreadable. Sometimes it feels like this: http://corsairmediaservices.com/images/blog/bad-interface.jp...
Unfortunately, I haven't seen much improvement on the design side. While their company has invested in design resources, it feels like every designer is focused on a specific feature. When you group those features into a page (like merge requests) you get this jumbled mess.
Luckily after using the product for 3 years you start to have muscle memory of where things are but overall the design severely lags Github's.
"While their company has invested in design resources, it feels like every designer is focused on a specific feature. When you group those features into a page (like merge requests) you get this jumbled mess."
Thank you for the honest and direct feedback. We work together as a team and try to keep a holistic view even as we iterate on specific features. We know we can do better here and are actively working to make things consistent and understandable. We have opened several epics dedicated to this, you can see them here: https://gitlab.com/groups/gitlab-org/-/epics?scope=all&utf8=...
GitHub might be closed source, but it doesn't matter that much in practice. The code is not where most of its value is. The value is mostly in large community of users and reliable infrastructure.
Also, for me, GitLab's heavy reliance on JavaScript disqualifies it as a serious competitor to GitHub.
>Also, for me, GitLab's heavy reliance on JavaScript disqualifies it as a serious competitor to GitHub.
How in the world using JavaScript makes a project less serious. Although Gitlab use mostly use Ruby and Go. JavaScript is used for Web UI only like most companies.
I'm guessing the key word here is reliance. Some of us are fine with sites that use JavaScript, but only if the core functionality still works when scripts are blocked.
If you're wondering why someone would feel this way, the reasons vary from person to person, but perhaps the best reason is that a site that becomes nonfunctional when client-side scripts are disabled puts itself at odds with basic web security precautions.
That JavaScript isn't a serious programming language is a decade old understanding of the ecosystem and a disregard to the current day use of the language. A lot has happened in the last decade to make JavaScript "more serious" like NodeJS, TypeScript, and React. Even on the back-end, it's becoming more and more popular.
Eh, I think they were talking about the UX for people that have JavaScript disabled, not about the JavaScript language.
To me it sounded more like "I can browse GitHub without JavaScript just fine, but GitLab doesn't even display the README, even though I'm using the exact same settings".
Sounds like a problem with the user, not a problem with the UX. There should be no expectation that a website be functional when JavaScript has been intentionally disabled. You never see people who have disable CSS complaining that a website looks terrible without any styling applied, why do people still expect that a website function to some arbitrary degree when they’re the ones who have deliberately disabled its functionality?
It's both a user and UX problem. It's like when some part of a website's functionality is only available through Flash.
You know some (nowadays "most") people will have it disabled, and you're indirectly saying that you don't care about those users at all, which is not a bad thing.
It's the users' fault for not enabling Flash, sure, because they're not your target audience. But whether you provide a good UX for your non-target audience or not, it's your fault. And again, you're not obligated to accommodate those users, but that doesn't change the fact the you could give them a better UX but chose not to.
So, from this you can know that the users GitLab cares about are only a subset of the users GitHub cares about.
The advantage of a large community of users goes away as soon as you can implement federated identity, which can be as simple as OpenID. (Heh, OK, could probably be simpler than that, too.)
I've been using GitLab for a while.
And for me this looks like a PR move in the light of Github being bought by Microsoft.
EDIT: did not notice post was from 2016. Has nothing to do with Microsoft buying Github.
GitLab (not self-hosted) has constant deployment/stability issues. They do an update and sometimes GitLab is down for several hours.
This is not a huge deal for me, as I can just push my code later.
But the main concern I have is that recently they've just removed some free features in order (I guess) to force people to pay.
Features removed (the only once I've noticed):
- Merge requests: squash commits feature
- Push rules: make sure users do not push commits with non-Gitlab user emails.
- Protected branches: allow certain users to push/merge, not a whole role.
There was no email notifying about this changes, it just happened that I've created a new repo, and then noticed some stuff missing.
EDIT:
Since this got traction. I've started digging to see the differences between repos.
I have one group that is under "Early Adopter" plan, which has all the features mentioned.
Recently I've created a new group, which went under "Free" plan, and this group does not have features mentioned.
I wonder why the "Early Adopter" plan is not carried over to my new group.
> For existing users on the Free plan, we've created a special Early Adopter Plan for you. This plan has all of the existing features available in our Silver plan, with the exception of additional CI minutes or premium support. Any group or user account created before September 1st will be put onto this plan for a year for free. While we will not add new paid features to this plan, you'll continue to enjoy powerful features, like multi-project pipelines and canary deployments, for the next year. After 12 months, you will get rolled back to the Free plan. You can upgrade at any time.
Still personally feels like they're taking away free features, but just giving me a year to enjoy what I've already had.
Yes, I (submitter) do not work for GitLab, I just remembered this article while replying to a thread in one of the submissions about the acquisition, and thought it was interesting to read again in that light.
I can still do this as a free user. It looks like you just have to do it within the "edit" form, or when you create the merge request. https://i.imgur.com/70k53xr.png
I don't use the last two so I can't speak about those, though.
Note: I'm assuming you mean the hosted version at gitlab.com
Yeah, well...Gitlab is also built on libgit2 and rugged. The core libraries that power Gitlab’s interaction with Git? Those are open-source projects written by Github.
I did some work with libgit and rugged and said to myself that I was shocked Github was open-sourcing such a core competency...these libraries make it pretty easy for an average web programmer to build their own Github.
Github has been pretty damned generous to the open-source community, and it’s sad to see a fast-follower project use this kind of rhetoric successfully.
"""GitHub very actively contributes to open source themselves, including many contributions to Git and Ruby on Rails, and releasing libraries and applications like libgit2, Atom, and Hubot. Also note that we build GitLab with software that GitHub open sourced such as libgit2."""
The article is profoundly dishonest. They've built their whole company on open source code provided by the "closed source company" Github.
Here's the truth that so many seem to be in denial about: it is "closed source" companies like Google, Microsoft, Facebook, IBM and Oracle that produce the overwhelming majority of high-quality open source software. It's not even close. Over the last ten years these companies have given away billions of dollars in value in the form of open source software. These companies have done much more for open-source than GitLab and their buggy Javascript ever will.
...and it has a yearly donation drive. Okay for a community project (in the sense of "steward to public data"), not okay for a service provider. And a service provider needs to cover costs - e.g. by running a customized, better version of the OS product.
I think the people having these reactions aren't going fall for any PR stunts, no matter how grandiose. Even with the OSS they have, they're constantly undermining the spirit with opt-out telemetry, some even forced, in fundamental tools, and they are unapologetic about it. Even if they announced they were taking all that out, I wouldn't trust it to last, or that they wouldn't use that PR to leverage something even worse.
I've hated Microsoft for much of my life. I try not to make it a religious issue, but I'm worried about github. If they open sourced it I would no longer be worried.
They won't, though. They won't open Windows or Office either.
Why are people worried about Github? I mean, I use it and I like it, but I also have local copies of my code. If I stop liking Github, I feel pretty confident I could set up another workflow with Gitlab or even self-hosted git if I had to.
I mean--I used to use Skype all the time, but now I use something else. If MS breaks Github, there are options.
I wouldn't know how to migrate that between any systems. Maybe what's missing -- more than open source code -- is open formats for that sort of data? Like we just need a button to export that from github more than I think I'd need Github's server code.
Github is very unlikely to be a massive cash cow, unlike those other two. And even if they make it Open Source, who's going to compete with them? If Microsoft revs up the enterprise sales machine, there's only a handful of companies that can compete with them. And none of those will want to build a company on FOSS tech owned by a direct rival.
This is not a case of Red Hat EL vs Oracle Linux. Microsoft has a mean enterprise sales machine and it's huge. All those Microsoft sales folks want to hear from a competitor is "oh, by the way, our enterprise SCM offering is based on a Microsoft solution".
It's a shame code doesn't drop into the public domain after a decade...
But windows must be a colossal mass of who knows what at this point. I'd be embarrassed to open that cupboard for all and sundry to inspect. And likely a double edged sword in terms of security.
I remember a quote by a philosopher that said corporations should be dissolved after 10 years.
The idea - in any form - really does make one think, and kinda hard. The world would indeed be a very different place. I wonder what kind of innovation such a move would stifle?
Hmm, maybe superscalar/moonshot-level competition would be defined by and buoyed by 10-year disclosure! That would be kinda fun...!
Regarding Windows, those who poked around the Win2K and WinNT leaks regarded the code as reasonably well-written, so there's that. The Windows Research Kernel is somewhat more straightforwardly accessible; "wrk-1.2" will find innumerable copies rehosted on GitHub.
Regarding security, yes, it would be a payday loan. The NSA tool leak had some pretty old things in it that still did a fair chunk of damage.
Satya's choice of CEO for GitHub in Nat Friedman seems to point to a previous precedent on the matter. When Microsoft acquihired Nat Friedman he was the CEO of Xamarin, and Microsoft open sourced all of Xamarin within several months of the acquisition.
Obviously there are few if any guarantees here, but if Satya is going to find a way to open source all or most of GitHub, placing Nat Friedman in charge is certainly a sign.
Oh, it's in those two footnotes: Apple pre-announced that they were going to open source Swift, and then they did, and they did a great job and there was much rejoicing.
But there was some skepticism, because Steve Jobs had previously announced they were going to open-source their FaceTime technology so non-Apple platforms could adopt it and it could become a cross-platform communications system... but they never did.
Both announcements got people excited, though.
EDIT: Okay, my bad; I stand corrected. I remembered him saying that in 2012, but I googled it after lowtolerance disputed my claim, and what he actually said was, "We’re going to the standards bodies, starting tomorrow, and we’re going to make FaceTime an open industry standard.”
Apple never said anything about open-sourcing FaceTime. Jobs said they were going to make it an open standard. That’s very different, but it doesn’t matter because they didn’t do that, either. My understanding is that patents got in the way of the plan, but according to engineers at Apple, there was never any actual “plan” to do this. Jobs’ introduction of FaceTime was the first they’d heard about making FaceTime an open standard, too.
Apple open-sourcing Swift doesn't really do much though. Sure, Swift is a nice language but allowing IBM or others to start using Swift for making mainframe or web frameworks doesn't make a difference to Apple from a sales perspective. Now if they either open-sourced iOS or started a program by which OEMs could sell devices with iOS then that would be huge -- but that's not going to happen because Apple has been rather up-front that they are a consumer electronics company now so anything that diminishes their ability to sell hardware is counter to their prime directive. Google, on the other hand, makes money by having more information to index, analyze, and develop marketing profiles. They would be just fine with Microsoft putting Android on a Surface Phone as long as hooks to Google Cloud products remained... otherwise they wouldn't be able to call it Android (eg: why Amazon's Fire devices don't officially run Android even though 99% of the APIs are the same).
I don't think taking the wind out of critic's sails is worth over 7 billion. In fact, if whatever MS does is not worth at least 7 billion, then the shareholders are not going to be happy with Microsoft's executives.
Github is more a service than a software, you need tons of components to make it work (Elastic Search, a DB, haproxy or nginx, a lot of ruby, a few C/C++ libraries, a few python libraries, etc) it's a big beast.
I know they provide VM images for Github Enterprise, but IIRC, it was quite a big beast, requiring at least 16GB of RAM a fast CPU with many cores and a large disk last time I looked (3 or 4 years ago). And there was a lot of stuff in that image.
If I had the source code, I'm not sure I would be able to properly deploy it and maintain, even with some documentation. Too many moving parts. And I feel a lot of sysadmins would be in the same situation.
There are some services that open source there code. Travis-ci is the one I've in mind, yet I don't know of any deployments outside of travis-ci.
I too wonder if they would. It would be for the best if they did. They have a lot more to gain if they did and GitHub already has an enterprise self hosted option so I assume its already generic enough code wise to be redeployed anywhere.
a.) A huge part of GitHub's value is their default-ness. Anybody can build a website UI to git, although it is a lot of engineering work. Gitlab and Bitbucket are somewhere in that equation.
People uneasy about MS are all over the Internet talking about moving off of GitHub today. MS announcing that they are open-sourcing GitHub would instantly evaporate a lot of that doubt/resentment/NOFUCKYUOM$!!!!URRRGH.
b.) They could also open-core it, not totally open-source it, just like GitLab. Thereby keeping basically all of the more-lucrative enterprise business (that's probably way more comfortable with Microsoft anyway).
The main thing github gives microsoft is a pool of people it can try to push into Azure. Azure means that Microsoft no longer needs to care whether you develop apps for windows or linux or mac, or the web. As long as you make and deploy apps to Azure, you can use whatever stack you like.
For that reason, they have a strong incentive to keep people happy and on github, and they have a huge incentive to make it easy to deploy from github to azure with a single click or something similar.
It ties in nicely with the other open source (compatible) solutions they're trying to make money off of right now. The money is in GitHub Enterprise, and in how they can roll the product's workflow, issue management, and huge developer base seamlessly into their cloud offerings.
They don't lose any customers by open sourcing it. Nobody's going to use GitLab if they can just download GitHub, and companies will still pay for Enterprise because they still need support for their massive investment in GitHub's extended features.
The code maintainers that (now) want an open source hosting solution are leaving anyway. Microsoft also already owns a git-as-a-service solution in Visual Studio Team Services. My take is this is another way to get closer to people and orgs that will probably need cloud services of some sort, so make it as easy as possible to choose Azure. I think the OP makes an excellent suggestion, as a dev and a (infinitesimally small) shareholder I would be for it 100%.
I don't fully understand what you mean, please help me understand. Why would the generic term "code hosting" change to two brand names, and what does S3 have to do with all this?
By "code hosting" I mean that most places say "we support code hosting" but in reality their systems are only setup to work with/use github.com; the domain at the very least, and possibly the API on top of that. Now places are going to have to generalize a bit, at the very least by thinking of multiple domains.
There are undoubtedly going to be tiny differences between how gitlab handles things; those will surface over time. They probably won't make headlines, but they'll cause tiny localized headaches.
As for S3, I mean that the GitHub API (issue tracking, etc) is all non-git proprietary stuff. GitLab should copy that so people's workflows Just Work™. I have no idea if they've already done this, I'm sure they've considered it.
I have a major question about self hosted software like GitLab.
As many of you know, I am a huge proponent of open source and the collaboration economy.
I have put my money where my mouth is by reinvesting over half a million dollars from our startup company’s revenues into building the open source Qbix Platform, in the hopes of leading a revolution in all SAAS software, including the stuff where you need profiles, permissions, collaboration, notifications, realtime and social features on top of whatever you’re doing (git, editing a document, planning a trip).
Having built v1.0 recently, we began making backwards compatible changes to eventually turn the Federated model (communities = landlords) into a permissionless, end to end encrypted and secure model for everyone. It may take us another year:
BUT here is the question I have. What is the best way to UPDATE the software on the back end?
We want this thing to be secure. Discourse, Wordpress et al tell you about updates. That’s fine but how do you actually install them? By entering a password into a web interface you can get the web server user to replace all the core files of the installation? Isn’t that a vector for serious attacks?
All the attacker has to do is somehow get your password or password hash, and it’s game over. They can send the right request to do arbitrary code execution.
On the other hand, if a developer signs the new payload, this may mitigate things. It is what app stores do.
Still, how do back ends get upgraded in self hosted open source software? In SAAS you just don’t worry about it. This is one area where we could use some professional consulting :)
> BUT here is the question I have. What is the best way to UPDATE the software on the back end?
GitLab has the Omnibus packages. We build it for few major linux distributions and you can get the package from here: https://packages.gitlab.com/gitlab/gitlab-ce (you can also add it as debian/rpm source, see installation).
Upgrading is a matter of `apt update && apt upgrade` and checking for eventual configuration changes/deprecations in `/etc/gitlab/gitlab.rb`. If you read the release note posts, upgrade instructions are always on the bottom of the page.
All packages are GPG signed, so you can check for tempering. If you don't trust our builds, you can fork our omnibus-gitlab repo and trigger a build on your own hardware.
The only thing keeping Us from deploying access to Gitlab to the rest of the Organization is the lack of multiple file upload in the Web UI
Gitlab is really an amazing piece of software and We would like to have all the Organization on Board to handle Everything alas Redmine
It's the only thing missing in our criteria or use case taking into account that Gitlab is not a software for the whole Organization and more oriented to software development.
But adding a pair of things, it can be of use to all the departments in an Org
I've used it a very little. They've unfortunately copied the terrible workflow around github "pull requests", except made it even more confusing. The rest of it looks reasonable.
Wondering whether Microsoft has ever acquired any major open source project/product? from memory, minecraft/skype/github are all closed source stuff. It is also pretty amazing that Microsoft still refuses to open source any of its major products like Windows, Office, SQL Server or even Visual Studio - even when most of them have better open source alternatives.
If you look, the most embarrassing thing about Microsoft is that they pretend to be embracing open source under the new CEO, yet their entire WSL joke aimed for running linux apps on Windows is not open sourced. There are numerous WSL bugs, there is an official github repo for WSL there but there is no WSL source code you can read.
Maybe for Microsoft Open Source is still considered as a cancer?
Xamarin was also closed source and $1000 per platform per year (my employer at the time was paying $3k/year for the mobile devs to churn out iOS, Android, and Windows mobile apps from one Xamarin code bas). Immediately after being acquired by Xamarin the price was dropped to zero and it was announced that the entire Xamarin platform would be open-sourced (and this promised was fulfilled less than a year later).
This is not the Microsoft of Balmer where Linux was called a cancer. Under Satya Nadella the company is serious about providing the best office and cloud products regardless of what OS you use.
I was excited when I heard about WSL. Of course I found the name confusing, Windows Subsystem for Linux, isn't it a Linux Subsystem for Windows?
Then I realized that the COM ports are not routed into WSL. I need USB-to-RS232 converters to interface with our hardware. Of course, there is a web of contradictory blog posts and announcements when that feature will be finished - one post stating if I upgrade to Creators Update MVP Edition 120394782034785023845 Build 23904782390475 or later, I can use the feature! It took me 1h to find out what version and build of Windows I had, and 1 hour to install the new Creators Update, and 1h to reinstall WSL - and guess what, the feature was absent. What a clusterfuck.
I think MS is trying to both support and not-support WSL, in the sense of building it out enough to support their internal agenda, but not where complex SLAs are needed.
So, COM port access... I don't see that being hard; I envision it as boring API-wiring, since RS232 has been standard since before Windows and Linux existed, so it's likely the code's generally similar.
But in terms of high-level support, RS232 is primarily industrial-class, with a long tail of developers and workstation uses trailing after it. Industrial immediately means complex SLAs and - nooooo - industry certifications and the like. I don't think MS wants to invest that much into WSL.
For a company whose existence is largely due to selling software licenses, is this surprising?
That being said, I wouldn't be surprised to see a larger open sourcing effort at Microsoft in the coming months/years. Let's hope they do it better than Oracle.
If .NET, Powershell, VS Code are any indicators, they're getting really, really good at Open Source stuff. They run a really tight ship and all it takes is to point it in the right direction. Just look at this: https://github.com/Microsoft/vscode/issues/50477
It's basically their May 2017 release plan for VS Code 1.24. Fully public.
My guess is that they're preparing some major announcements, I think at some point Windows will be Open Sourced and the consumer editions will be free. Basically we'll all be beta testers for the Pro and Enterprise editions. Which is not half bad, if they ramp up QA a bit.
My guess is upwards of 90% of users would not use features in MSOffice not available in Open/LibreOffice. Most organizations are "creatively encouraged" (ie. forced) to use MSOffice.
Similar for SQLServer - Windows shops kind of default to it.
I would put anything made by Jetbrains in front of VS in the imaginary queue ordering the products in the same ball-park, not just close. Some of their products (Intellij Idea Community, Pycharm Community) are open source.
Well, it certainly will be for the next few days as GitLab wraps their minds around the new scale they're working with (<1k API calls -> 20k API calls \o/)
GitLab is self hosted, and I don't know what you mean by slow? Just see what the bottleneck is (disk/memory/cpu) and use a higher powered instance.
If you mean GitLab Hosted is slow? Well it's probably getting hosed and that's understandable. I did find that GitHub was slower for us though and had far less features.
I have this one static page that uses their API to list all pipelines across all projects. Like 50 API calls at once. It's usually terribly slow (1-2 mins to finish, sometimes more).
Today it's the fastest it's ever been. <1 second to load.
The marketing department of Gitlab seems especially active on HN lately. Why is this article being revived? This was already talked about two years ago.
You know what is also open source? Git. I am not sure what the value is for having an open source clone of Github. Repositories can be changed in seconds thanks to git. So unless people are actually afraid of Microsoft reading private repos and can’t afford Github Enterprise, than I fail to understand why it matters that Gitlab is “open core.” Site reliability is far more important. I have never cared to look at Github source code. What for? It’s a utility for me like Dropbox or iCloud. GitHub and Github Enterprise have always worked great for me; as long as that continues to happen, why would someone like me care about Gitlab? Can anyone provide a specific benefit of Gitlab over Github? At the end of the day, what’s the value proposition?
I agree it isn't the best article to revive when there are fresh articles on the major news. But I don't think it's "the marketing department of Gitlab". It's just that a bunch of users want to talk about Gitlab and open-source generally in the context of the news. Same reason https://news.ycombinator.com/item?id=17223116 was a huge thread yesterday.
I have 1400+ projects and more than 700 users in 3 geographic locations using my server. backups with ruby fog libraries included in gitlab are taken once a night and im expanding the CI presence to include a kubernetes cluster. In short, Gitlab is worth it if you have the resources to deploy.
If not, gitea is a great alternative as well. https://gitea.io