Ben points out that the scenario I'm talking about is not the same one
Jim originally proposed, so please ignore what I just wrote while we
think about it.
> Ben Collins-Sussman <email@example.com> writes:
> > Erg, I disagree. You've hit on the central issue: copying a file to B
> > and renaming it back again *are* explicit actions -- they're
> > full-fledged, intentional "local changes". As a user, I would be
> > mighty pissed if Subversion decided that it was ok to ignore my local
> > modifications and assume I'm not up-to-date -- "oh, that copy and
> > rename? That didn't really do anything useful, so I'm just going to
> > pretend they never happened."
> Subversion isn't ignoring them, it's just reducing them to their
> smallest total effect: namely, that nothing happened. If you rename A
> to new file B, and then rename B back to A, and never committed any of
> it, then nothing happened. The user may have a memory of something
> having happened, but how would its result be any different from the
> situation that Subversion reduced it to?
> > Maybe it's a matter of personal taste, but I hate it when software
> > tries to second-guess a user's intentions. Local mods are local mods,
> > whether they're silly or not; in Jim's example, the working copy is
> > *already* up-to-date. Why? Because nobody has changed the
> > repository -- end of story.
> > (Girding asbestos...!)
> If there's guessing involved, then I agree -- Subversion shouldn't
> make fuzzy guesses except when you ask it to. But this isn't a guess.
> What happened is clear and unambiguous. To put it another way, the
> effect of each operation was fully understood by Subversion, and so
> the "sum" of the operations is also fully understood. Why shouldn't
> it make use of that knowledge?
Received on Sat Oct 21 14:36:07 2006