For context, Matt Mackall is the BDFL of the mercurial version control system.
EDIT: I guess benevolent dictator for the next 10 months or so? Not so much for life…
Is anyone using Mercurial instead of Git? I thought about switching to Mercurial once, but now it seems the project is slowly dying. Are there benefits?
Mercurial development is not dead at all:
The userbase is dwindling, but the development, if anything, is speeding up.
I use it almost exclusively for my personal projects.
I’ve found that Mercurial’s plugin system lets you build any workflow you want straight into source control. I also don’t think Mercurial is dying off, just that Github has really pushed Git up and nobody has tried to do something similar for Mercurial.
There’s a couple of people at bitbucket who care about really pushing the envelope with what Mercurial can do. Sean Farley is rolling out Evolve for select bitbucket beta-testers upon request.
Any public information on this change?
I don’t think so, no. Feel free to stop by the #bitbucket or #mercurial channels on freenode to ask questions.
That’s good to hear. I use Mercurial on all my personal projects and strongly prefer it to Git, but reading the blog posts and announcements from Atlassian, it’s really felt like the development velocity there has much more been on the Git side of Bitbucket.
I started using Mercurial for work, and have since grown to prefer it over Git. In large part because of it’s extensibility, but also ease of use. Mercurial makes more conceptual sense to me and is easy to figure out from the cli/help alone. I rarely ever find myself Googling how to do something.
I still like Git though, and it’s likely better for people who don’t like tinkering with their workflows.
Lots of people, including some big names (e.g., Facebook). I find git’s merging more reliable, but prefer hg’s CLI. They both get the job done.
I’d love to know about cases where you find git’s merging to be more reliable. Samples would be awesome, so we can figure out what’s tripping you up.
It’s a known issue.
Sort of. It’s not a known issue that BidMerge (note that we’ve shipped BidMerge, which is an improvement over ConsensusMerge as a concept) produces worse results than Git. I really meant it when I said I’d appreciate examples, rather than handwaving. :)
I was using hg pre-3.0 (via Kiln). The problem that BidMerge is intended to solve is the problem which gave us so much trouble. I can’t speak to how well BidMerge would have fixed that, as the company is no longer in business.
Fair enough. It should be pretty well solved then. Thanks for responding!
It may well have technical advantages, but if you’re working on a project that other people will one day work on, I’d strongly urge you to use git. Being able to use a familiar tool will be far more valuable to other contributors. Look at e.g. Python, which chose mercurial years ago but has recently decided to migrate to git.
I am a Mercurial user in a quite narrow way: we used it to store code for one of our daemons. That choice was made by a co-developer that didn’t like my team’s existing VCS. I’ve not felt in any hurry to unwind that decision, and I guess I don’t feel any extra hurry now. But I won’t be signing us up for new uses of it either.
A tiny handful of open source projects I track work out of Mercurial. All of those also release tarballs, so I’ve rarely had to deal with it otherwise. I spend more time working with upstream via Subversion than I do Mercurial.