Can anyone here comment on atlassian's stash vs github's enterprise offering? It seems nice to have jira integration while github is clearly very popular in the open source world.
One of the differences is an insane difference in price. According to their two respective pricing pages a 500 user installation would be $12,000 with stash and $125,000 with GitHub::Fi.
Because I use Mercurial. They're going to wind up sacrificing their Mercurial business, where they're the market leader, in order to play second fiddle to github.
I love their stuff and have used bitbucket since prior to the acquisition, and I respect that they may have to do this for business reasons, but that doesn't mean I have to like it.
I wouldn't be surprised if their Mercurial business is stagnating because, let's face it, Git has won. With that in mind, it might not be such a bad thing if they sacrifice a market of shrinking value.
Git currently has a greater mindshare than mercurial, specially on HN and many open source circles. I don't think that is enough to consider that "git has won". There are big projects and big companies that use mercurial instead of git (e.g. Mozilla, Python, Facebook). Both git and mercurial are great and both can (and probably will) coexist for the foreseeable future.
If you only read proggit or HN you could think that everybody uses git. The reality is that there are many teams that use mercurial. There are many, many teams that are still using old style non distributed VCS. There are lots of teams that don't even use any VCS at all!
There are lots of people (myself included) who consider that git is needlessly complex, and that mercurial hits a better trade off between power and complexity. A lot of teams would be much better served by using mercurial instead of git. I know for example of several HW design teams that decided to use mercurial because they found it much easier to use.
In addition, I don't think that git or mercurial, as they currently stand, are the best we can do on a DVCS. There are plenty of areas where they both can improve (submodules / subrepos being one of them, but also better support for code review workflows, simpler / safer history rewriting, etc). Perhaps something else will come that will learn from git's and mercurial's mistakes and improve on what they offer?
What I mean by this small rant is that it would be foolish to consider that git has "won" and that there is nothing that can come and improve on it. I hope that the git devs don't think that way. I know that the mercurial devs definitely don't think that way, given the number of cool things that they are working on (e.g. with the evolve extension being my preferred one).
I'm in the same boat (I use git a lot at work because I work with the linux kernel but I prefer Mercurial for my own projects) but I feel the thing that may end up killing mercurial is git's superior branch support.
For local dev branches mercurial patch queues work great but working with multiple remote branches becomes a bit of a head ache. Generally you end up using several repositories instead of branches within the same repository.
That being said that was the state of the art a few years ago and I haven't followed Mercurial development closely so maybe it's not the recommended way of dealing with that anymore. Is there a way to properly delete a branch in mercurial these days?
Git branches is the only thing I could imagine a git user miss when switching to mercurial. And possibly staging, it's sometimes convenient (but can be emulated using extensions).
> Git currently has a greater mindshare than mercurial,
> specially on HN and many open source circles. I don't
> think that is enough to consider that "git has won".
I think it is enough, because a few years ago hg had bigger mindhare. Four years ago I talked about DCVS at a local PHP conference and the three systems I reviewed were Mercurial, Git and Bazaar. Mercurial was more polished and leading in mindshare at the time, but I bet on git, and turns out it was the right bet. Don't hear much of the Bazaar despite few really big projects using it.
All of the open source projects are on Github, absolutely. However, all of my closed-source professional work using git has been exclusively on Bitbucket and that was the case with a multitude of different clients.
Github isn't terribly appealing for an organization that has many, many repos. Bitbucket is.
We're a smallish agency who switched (when I came on board) to git last year.
Currently we have 67 repositories through bitbucket, and all free as we haven't reached the developer cap yet.
If you are going to have quite a few repositories, bitbucket wins out massively compared to per repository pricing that github offers which is good for small companies who are working on a single project.
We do have excellent support for Mercurial in Bitbucket, Bamboo (which I am Product Manager of) and Fisheye.
I can't speak for Stash but there is a lot of work going into it right now. It's still a relatively new product and both Git & Mercurial are important parts of many teams workflows, so we will have to wait and see.
What would be really existing is a venture that aims to teach Git to a "niche" like writers, liberal-arts students or an area outside that of programming.
I think that could be pretty exciting. They could basically do it without mentioning Git overmuch.
That's kind of what I'm trying to do with a project I'm working on [1]. Everything is backed by git repositories, and the interface allows you to work with git (branch, commit, fork, merge, etc) without actually knowing any git commands. Essentially, git is masked beneath the surface, but you get all of the power of git for versioning and collaborating.
