Bazaar still below the radar when evaluating VCS tools

[Andrew Bennetts]

Ben Finney wrote:
> "Stephen J. Turnbull" <stephen at xemacs.org> writes:
> 
> > The problem is that [Mercurial] will also trust the information given
> > implicitly by *not* using hg mv, hg cp, ....
> >
> > This is the big hole in the implementation of hg (and bzr, for that
> > matter) rename/copy tracking.
> 
> How is Bazaar's implementation flawed in this respect?
> 
> I've never had Bazaar silently accept filesystem changes as you describe
> above; it only records filesystem entry changes if explicitly told to do
> so. Am I wrong?

Well, you can explicitly ask bzr to guess what the filesystem changes
were post hoc with 'bzr mv --auto'.  But otherwise it doesn't guess.

I'm going to guess though: I think Stephen might be saying that none of
the systems automatically capture a 'mv foo bar' done in their working
trees, when ideally they would.  In principle you could have a daemon
with inotify (or equivalent on other platforms) watches on versioned
directories/files, and that reflects file renames into the VCS.  In
practice I suspect it's easier said than done (e.g. distinguishing real
renames from editors that save files by writing a temporary file then
renaming over the top of the old one), but perhaps someone will prove me
wrong :)

-Andrew.