I hesitate to say this because I actually have an anti-subversion rant in the works, but I think he makes a good point here.
For any develop who is at all serious about their craft, it's absolutely insane to go on using a crippled tool like subversion. However if you have a team of talentless monkey coders mixed with a generous helping of non-technical asset producers (designers, copywriters, etc), then modern VCS won't really give you any compelling benefit, so you are likely better off taking the better tooling.
In the long run Subversion would be better off if they just rewrote it on top of git plumbing, but barring that, it's not necessarily true that everyone has much to gain from modern VCSes.
I tried to like Subversion for quite a long while (any improvement on CVS would be good). In the end the main differences from CVS were. 1: a repository that (at the time) you could not delete "oops I checked in a DVD" from. 2: an Eclipse module that broke during most Eclipse updates.
When svnmerge.py inevitably yaks all over yet another branch we're trying to merge to a release candidate after a rebase or two, I'm only barely capable of figuring out how it got confused and how best to recover, and the rest of my team (experienced developers all) isn't even as good at it as I am. I hear 1.5 with merge tracking isn't that much better.
svn is an extremely leaky abstraction, and just not a reliable enough tool to unleash on the people you're describing. Fortunately we're switching Real Soon Now(tm).
Comments
I hesitate to say this because I actually have an anti-subversion rant in the works, but I think he makes a good point here.
For any develop who is at all serious about their craft, it's absolutely insane to go on using a crippled tool like subversion. However if you have a team of talentless monkey coders mixed with a generous helping of non-technical asset producers (designers, copywriters, etc), then modern VCS won't really give you any compelling benefit, so you are likely better off taking the better tooling.
In the long run Subversion would be better off if they just rewrote it on top of git plumbing, but barring that, it's not necessarily true that everyone has much to gain from modern VCSes.
I tried to like Subversion for quite a long while (any improvement on CVS would be good). In the end the main differences from CVS were. 1: a repository that (at the time) you could not delete "oops I checked in a DVD" from. 2: an Eclipse module that broke during most Eclipse updates.
When svnmerge.py inevitably yaks all over yet another branch we're trying to merge to a release candidate after a rebase or two, I'm only barely capable of figuring out how it got confused and how best to recover, and the rest of my team (experienced developers all) isn't even as good at it as I am. I hear 1.5 with merge tracking isn't that much better.
svn is an extremely leaky abstraction, and just not a reliable enough tool to unleash on the people you're describing. Fortunately we're switching Real Soon Now(tm).