On Sun, Aug 28, 2011 at 05:40:49PM +0200, Michael Sperber wrote:
OK, I now have the CVS repository massaged to the point where cvs2hg
will accept it. There's a basic decision to be made wrt. the
#1 I can convert everything to one big Mercurial repo. (About
#2 I can convert every individual package into a Mercurial repo, and
arrange them as subrepositories to an umbrella repo.
#1 has the advantage that it's easy to do: However, there are
import branches in CVS, which create a seriously mangled revision graph.
#2 is more attractive from a design perspective, and it's what
to do. It has a dowside, however:
There's still central content (the various build files etc.)
needs to be in sync with the invidual packages. Mercurial handles
this just fine, but I don't know how to create a consistent set of
repositories from CVS. If you want to actually build an old version
of a package, you'll have to look up the revision of the central stuff
that matches it manually. This won't be necessary for revisions after
I hardly know Mercurial at all, but one question: would strategy #2 cause
any fragmentation? I.e., are there any operations one can perform on a
single repository (perhaps some sort of searching) that would be awkward
or impractical on strategy #2 (requiring, maybe, a shell for loop)?
Cheers =8-} Mike
Friede, Völkerverständigung und überhaupt blabla
Alan Mackenzie (Nuremberg, Germany).
XEmacs-Beta mailing list