[Gnu-arch-users] Re: Which cscvs to use?

[Andreas Fuchs]

Today, Robert Collins <address@hidden> wrote:
> On Mon, 2003-09-15 at 17:21, Andreas Fuchs wrote:
>> Right. I could get yours to work out-of-the-box. All the branches
>> have one problem, though: When I try to import SBCL
>> (http://sbcl.sf.net/) from CVS, it does not group the first CVS
>> import into one changeset. Instead, cscvs commits every file as a
>> separate changeset, which results in patch-1 to patch-300-odd with
>> patch logs reading "initial import", each adding one file.
>
> Ah yeah. So what you want to do is 'import' the first changeset, and
> then 'commit' the remaining ones on the branch.

Actually, that is what I have done:

~/hack/to-arch/sbcl$ cscvs -i 1 ../export
~/hack/to-arch/sbcl$ cscvs -c 1:: ../export

This should do it, no? BTW, I'm missing an option to display which
changeset is the first in a given branch name. Currently, I see only the
branch names with a changeset count, as output by cscvs branch -B.

> And what you'd -really- like is an automagic version that imports the
> first changeset on a branch, if the branch is MAIN, otherwise it tags
> from the relevant changeset, and then commits from there on in.
> Possibly, you'd want to do that recursively, so you could start at an
> arbitrary changeset, and tell cscvs to make it all happen. The problem
> is though, that cscvs ids and arch patch-X's don't line up, so you
> need to look in the headers for CSCVS-id: MAIN-9334 or whatever. All
> good concepts.

Sounds good.

> For now though, I *think* I already suggested a recipe - import one
> changeset, followed by commits.

Well... /-:

Thanks for the hints,
-- 
Andreas Fuchs, <address@hidden>, address@hidden, antifuchs