On 12-Apr-2001, Larry Jones <[EMAIL PROTECTED]> wrote:
> Fergus Henderson writes:
> >
> > The reason that this broke things is that `cvs checkout'
> > was checking out inconsistent versions of different files.
> > After the import, some files -- those which we had not modified -- now
> > had revision 1.1.2 on the vendor branch, and only 1.1 on the main branch,
> > and `cvs update -A' retrieved revision 1.1.2, i.e. the vendor branch.
> > But for other files, their version on the main branch was 1.2, and on
> > the branch it was 1.1.2', and `cvs update -A' retrieved revision 1.2,
> > i.e. the main branch. So `cvs update -A' was checking out some files
> > from the main branch (based on gc_4_1 with our modifications)
> > and some from the vendor branch (gc_6_0_alpha6).
> > Trying to mix incompatible versions naturally broke things.
> >
> > Is this a bug in CVS?
>
> More of a design limitation than a bug.
Well, regardless of what we call it, it sucks :-(
> Just off the top of my head, I
> can't think of any good way to avoid this -- the merge is an integral
> part of the import, but there isn't any way for CVS to help with the
> merge if you don't actually do the import.
There ought to be a way of doing an import onto a branch,
rather than onto the main line. Doing the import would then
affect the branch, but would not affect the main line.
--
Fergus Henderson <[EMAIL PROTECTED]> | "I have always known that the pursuit
| of excellence is a lethal habit"
WWW: <http://www.cs.mu.oz.au/~fjh> | -- the last words of T. S. Garp.
_______________________________________________
Info-cvs mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/info-cvs
