Some of the responses here are kinda obnoxious. I helped kickstart the original letter and I'm very happy about this response.
GitHub more than anything has been a blackbox, and this was a very notable first step towards opening up. It should be encouraged, not shut down.
Privately (and some publicly), in the past few days, a lot of major open source projects were discussing a coordinated move away from GitHub. This likely puts that all on hold, but we'll see what changes GitHub makes and how people want to respond to it.
Again, I'm very happy about this response, as should everyone in this thread. We'll see in the coming weeks what it really means though.
Gitlab is a worthy alternative. It's only missing critical mass, which gives Github the upper hand in considerations such as "all my foss contributions are on there in one place and that's what companies look at when they offer me a job".
Hey, sytse - I know you're reading this. Have you guys thought about a way of including, in gitlab profiles, user info from Github in a non-intrusive way?
While you are here, I have a question. As far as I can tell neither your service nor Github nor Bitbucket have a way for a user to organize their repositories other than a list.
This gets annoying if one has many unrelated projects.
For instance, I have some chess projects, some math projects, some Warhammer Online projects, and so on. With the linear list approach if I want some semblance of order I need to resort to a kludge like prefixing the names with chess- and math- and WAR- and so on and using sort-by-name.
It would be a lot easier to deal with if I could have a chess folder and math folder and WAR folder and put the appropriate projects there. Even better would be multiple levels of folders, and allowing a project to be in more than one folder at a time.
In the project information, there is a field to enter a list of tags. I tried using that, with the idea that someone who wanted to see all my chess projects, say, could look for all projects I tagged "chess".
Two problems: (1) I could not figure out what these tags actually do other than show up on the project settings page, and (2) I could not find any help on them in the Gitlab documentation because searching for "tags" brings up a bazillion things about Git tags. Maybe these should be called "labels" instead of "tags"?
I see that there is a Groups feature, and I have not yet had a chance to delve into that to see if it can substitute for folders.
Unfortunately, it's only for teams right now, it seems. Additionally, (and this may only be a me concern) you can't include repos in more than one project, like in the case that you have a common framework you build between separate projects. I mentioned this to them (may have even had a support request), but haven't gotten any feedback.
Basically, it's much more rigid that I personally find useful, but it seems like their first step in the right direction.
[Edit] Seems kannonboy said the same thing, but I'll leave my opinions here. Interested to see if anyone else has used it and has any opinions.
There are two things that might be relevant to your request. In GitLab you can star repo's and show that, obviously this would mean only one list. And we would love to see nested groups https://gitlab.com/gitlab-org/gitlab-ce/issues/2772 although this would require you to move the projects under the chess namespace.
It seems like you guys are proceeding down the directories path to solve this issue-is there any reason that semantic tagging wasn't considered? It would appear to be a good fit and might solve the nesting issue, but I could have totally missed some conflicting requirements. (I admit I was flipping through the issue thread fairly quickly)
Definitely agree. I have a project that has iOS, server, and android reopos and it'd be nice to have there be some notion that they're related, maybe even have issues sharable? Though I guess that could get really messy.
If you're OK with forum support it is free (like Gmail). If you want email support you have to pay $9.99 per user per year. No plans for mercurial support.
As somebody who loves Gitlab... no, what it's missing is Github's reputation of lasting uptime reliability. If they get it to the point where people trust it to always be up to the same extent as Github, the rest will follow.
We're not there yet but uptime is a lot better than two months ago (we can deal with Azure now, PostgreSQL and Redis are HA). But file storage is still a single point of failure, we're looking into moving to Ceph https://gitlab.com/gitlab-com/operations/issues/1
That's one of the definitions of critical mass. Reputation is built up over time and usage. Gitlab's uptime reliability isn't particularly known or unknown, mostly because, compared to Github, it's not used enough for those issues to come up.
Try talking to a dev, ask them why they use Github instead of Gitlab. Are they going to tell you "Oh I absolutely would use Gitlab but their uptime is such a problem"? Or are they going to tell you their stuff's already on Github? Their friends/coworkers are on Github? Their employers look for their Github profile?
As a sidenote I would totally use MySpace instead of Facebook if not for their notable history of extended downtime...
I partly agree, but those that have done the research or already use Gitlab know that Gitlab's uptime has been problematic compared to Github's, which is a bigger issue for most companies than the community. As sytse correctly notes below, it's a lot better than it was a few months ago - but the pessimistic/risk-averse way to say that is "as recently as two months ago they were having major uptime issues". I have high hopes for Gitlab and I think they're doing great things, it's always great to see new competition in the space. And I applaud their transparency, it's refreshing! But it'll require some more "burn in" time before I consider it a reliable Github replacement, all questions of critical mass aside.
On the plus side, if the manage to fix their problems at some point people will start trusting them to be up. It is not set in stone that they have worse uptime.
Absolutely, and I do think most of their problems have been fixed. So it's just a matter of time - the longer they are stable, the better their reputation will become.
> Try talking to a dev, ask them why they use Github instead of Gitlab. Are they going to tell you "Oh I absolutely would use Gitlab but their uptime is such a problem"? Or are they going to tell you their stuff's already on Github? Their friends/coworkers are on Github? Their employers look for their Github profile?
Just as an alternative point of view, a shop I knew used self-hosted gitlab in preference to github, because they were in China. That was actually how I learned about gitlab.
FYI - We have been running GitLab internally for over a year, we have not had a single crash or unscheduled down time, not even once. The only few things we had was that prior to about 6 months ago we noticed a lot of regressions after a major upgrade, i.e. small things not working but they weren't show stoppers, I spoke with a few people at GL about it and they were fantastic, they immediately set off a program to improve testing quality and in the last 3 months we've upgraded the day each new release has come out (weekly now I think?) and haven't had any problems that we've noticed.
(Well, there are quite a few weird UI decisions that also pushed me away, but they're minor in the grand scheme of things.)
EDIT: It wasn't my problem per se, but one of my users who had a problem downloading things for at least "several" hours. As this project isn't very popular to start with that's a comparatively huge loss.
Critical mass is hugely important, but it can also be surprisingly fickle. I remember when MySpace hit the tipping point and it seemed like Friendster became a ghost town overnight. A few years later when Facebook opened up outside of colleges, MySpace turned into a dead zone.
It's possible for the same thing to happen with GitHub, though Facebook's continued entrenchment is also a signal that it may not.
Facebook is entrenched still, because they deliberately tried to avoid the mistakes that MySpace and Friendster (et al) made. They grew and expanded, changed their feature set to ensure that Facebook had everything your average person needed. GitHub haven't been doing that so much, but hopefully this response is the beginning of a change in that direction
gitlab doesn't have the same appeal to me for browsing projects... at least not on the surface.
the amazing thing about github is that you can go to github.com without signing up and search a multitude of projects and their forks. that's awesome. it's a level playing field.
Is this not available to view signed out? https://gitlab.com/explore if not I think that's something Gitlab should make visible to anyone. But the problem here isn't that Gitlab lacks this it's that no one is using it to make the explore page interesting.
Their pricing model is different, which I think keeps the base feature-set neutered. They try to up-sell you to JIRA and Confluence, which disincentives them from improving the base tools, IMO.
Yes, the pricing model is different with the goal of making Bitbucket affordable for individual developers and small teams, no matter whether they work on public or private projects.
Do we want people to use JIRA? If they require more than simple issue tracking, absolutely. But that doesn't mean that we aren't improving the base tools.
Thanks! I'm really new to it and I'm using it for basic mirroring for now. But the UI is really enjoyable and I've liked the native mirror support in GitLab.
And yet, look at this blog post from 2 days ago: https://github.com/blog/2110-migrate-your-code-with-the-gith... still offering to migrate your existing SVN or Mercurial repositories to GitHub, without any mention of all the other data that are typical of a software project like issues. Should I migrate my Trac or Bitbucket issues manually to GitHub?
GitHub more than anything has been a blackbox, and this was a very notable first step towards opening up. It should be encouraged, not shut down.
Privately (and some publicly), in the past few days, a lot of major open source projects were discussing a coordinated move away from GitHub. This likely puts that all on hold, but we'll see what changes GitHub makes and how people want to respond to it.
Again, I'm very happy about this response, as should everyone in this thread. We'll see in the coming weeks what it really means though.