You can no longer afford not to take Git seriously

This post is more than 11 years old.

Posted at 06:00 on 08 June 2012

There is a blog post about Git that I’ve been wanting to write for a while, but I never got round to. In this, I would have expressed concerns that by being unnecessarily hard to use (with a command-line centric culture, borderline incomprehensible documentation, and some surprising, unintuitive behaviours), and by occupying a dominant position within the DVCS scene, Git was putting 9-5 developers off distributed source control altogether and scoring one own goal after another in the battle to displace Subversion.

That post will never get written. In the light of the results of this year’s Eclipse Community Survey, released yesterday evening, it would be patent nonsense. Those fears have been resoundingly shown to be unfounded.

Git’s market share, industry-wide, is now 27.6%.

Twenty. Seven. Point. Six. Percent.


That’s more than I expected in the best-case scenario. Based on its growth rate up to now (it scored 12.8% last year and 6.8% the year before), I was expecting Git to score somewhere between 16% and 24% or so, depending on how well it is being received in corporate environments. If my thesis were correct, I’d have expected it to score at the lower end of this range. I thought that 25% was the maximum it could possibly score in the best case, and even then, that was highly unlikely.

(The news is pretty bleak for Mercurial, by contrast: it has dropped from 4.6% to 2.6%. Subversion is still at number 1 for now, but down from 51% to 46%. This is a smaller drop than I expected, but no doubt it’s being shored up by a trickle of late adopters migrating to it away from CVS. Nevertheless, it looks almost certain to lose its number 1 position within the next 12 months.)

Clearly this result is a game changer.

It means that you can no longer dismiss Git’s mindshare as hype.

It means that, as a Subversion or TFS vendor, you can no longer plausibly claim that Git is only suitable for open source development and hobby projects, and not for corporate and enterprise environments.

It means that, as a recruiter, if you are not using Git, you will find yourself facing an increasing number of competitors who are, and a decreasing pool of candidates who are willing to tolerate the alternatives.

It means that, as a developer, if you’re not already thoroughly familiar with Git, you had better learn it. Now.

It means that, as a diehard Mercurial fan, I have finally had to concede that Git has won.