Such feature comparisons can be misleading (I'm not claiming one way or another for this Wikipedia entry, since I have deliberately not read it) because it's easy to dismiss a core feature of a product you don't like as "mere sugar" and not include it, while taking a fairly minor feature (or a very narrow definition of that feature) from another product you do like and add a whole column with a single check mark in it.
It's also possible to introduce misleading comparison results by doing an apples / oranges comparison where you compare your knowledge of one system with your ignorance of another, or you check features from a bullet point list for one system off against the rather disorganised documentation of the other, leading you to miss features.
The Shlomi Fish "Better SCM" site for example is very clear that Git won't do a merge across a rename. It even has a citation for this claim. But as a Git user who has actually done a merge across a rename I know it works just fine, and anyone familiar with Git's internals will guess immediately why. Yet probably there is no line of documentation on the Git site or elsewhere that I can quote to justify adding a "Yes" to the comparison.
Posted Jan 6, 2009 1:50 UTC (Tue) by tao (guest, #17563)
[Link]
Sure, I forgot to write a disclaimer about Wikipedia, and comparisons in general, but my point wasn't to claim that git (which I just picked as an example) is better than darcs, I just wanted to obviate that the post I was commenting on had a totally unsubstantiated claim.
The GNOME DVCS survey
Posted Jan 6, 2009 2:41 UTC (Tue) by dlang (✭ supporter ✭, #313)
[Link]
off topic, but I posted your comment to the git mailing list and the response I got (about 5 min later) was
quote:
I think it's documented in Documentation/merge-strategies.txt, under the
'recursive' merge strategy:
"Additionally this can detect and handle merges involving renames."