On 10/15/05, Miles Bader <[EMAIL PROTECTED]> wrote: > I maintain a bunch of Emacs-related trees in arch. However they are > all still officially maintained in CVS, so the majority of changes > take place in CVS land, from which I merge. > > Arch works _brilliantly_ in this scenario, precisely because taglines > _do_ survive the out-of-arch periods. If a tagline-bearing file gets > renamed in CVS, I still get precise rename info when I merge from CVS.
Now, that's cool! But I'd need to add taglines upstream, right? If that's the case, it works for you because you are the upstream maintainer. I don't see myself getting an ok from random upstream maintaners to add taglines liberally to their trees so that I can run my private branches ;-) > For files which cannot have a tagline added for some reason, my > merge-from-CVS scripts do something similar to what you say GIT does > -- they use "file similarity" to make rename judgements. Interesting! Are they available anywhere? I would have killed for an effective way of doing that. > The point being, having precise rename info doesn't preclude using > other methods of rename detection when the former isn't available for > some reason -- but when it _is_ available, you can do much better. > It's a pure win. > > [Arch's "identity based" (as opposed to "history based") rename > detection is a rather nice fit with such methods, actually.] Fair enough. I can see that some tricks that I missed could have made my Arch life easier. Oh, well. Had I only known. Still, don't jump to my throat if the (l)user in me says that all that burden on the scm user just to track renames seems all wrong. cheers, martin
_______________________________________________ Gnu-arch-users mailing list Gnu-arch-users@gnu.org http://lists.gnu.org/mailman/listinfo/gnu-arch-users GNU arch home page: http://savannah.gnu.org/projects/gnu-arch/
