3 ms·
Git is absolutely terrible. Let’s just get that out of the way. But it ticks the two boxes - powerful enough (to e.g do proper merging, which cvs and svn neve
by alkonaut 9y ago
Git is absolutely terrible. Let’s just get that out of the way.
But it ticks the two boxes
- powerful enough (to e.g do proper merging, which cvs and svn never could)
- tooling support, meaning it’s supported out of the box in bug trackers, build systems etc.
All others (hg, perforce, svn, cvs, pijul, fossil, ...) fail one or both of the above.
Now, I hope that one day git will be replaced by something nicer. But for the time being it’s what we’re stuck with.
My pet peeve: sequential revision numbers. Why not have an arbitrary numbering of commits?? Saying “I have bug X in rev 1234 but it’s not in 1230” is fantastically powerful compared to “I have the bug in a1b34h but not in 3ae452”. These would be a sequence for a particular centralized branch - typically mainserver/master.
Because this is the reality: it’s distributed version control but we almost all use centralized version control. This is also why it’s so odd that Git LFS took years to make it into git. Why would I want all past revisions of a binary (Yes, binaries must often be in version control, whether anyone thinks it’s a bad idea or not)?
- russdill 9y agoThat's why most people tag their builds with git describe. It can easily tell you the most recent tag in the current branch and how many commits away from it the build is, for instance v1.5-15-a55325.