On Wed, Nov 17, 2010 at 7:59 PM, Nolan Darilek <[email protected]>wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Figured out what the issue is.. Actually, I think it's two-fold.
>
> 1. The Fossil clone of my repository lacks a tag made after my final
> commit. IOW, I made a commit, then ran:
>
> git tag 0.9.0rc2
>
> But that tag is nowhere to be found. It shows up in my Git repository,
> however.
>

Unfortunately, this seems to be a fault in Git not Fossil.  The 0.9.0rc2 tag
is not appearing in the fast-export output, which you can see for yourself
using:

     git fast-export --all | grep 0.9.0rc2

We're finding that "git fast-export" has some subtle bugs - a fact that has
been confirmed by the maintainers.  Some of the issues we've been able to
work around.  But if "git fast-export" isn't telling us anything at all
about the tag, it isn't clear what Fossil can do to correct the problem.
Sorry.

Is there perhaps some option to git-fast-export that we should be using in
order to get better tag information?

If you run the output of git-fast-export back into git-fast-import, does the
new Git repository have the 0.9.0rc2 tag?


>
> 2. I recently made another git tag, 0.9.0rc1. I said that my master
> branch was missing commits from mid March? That'd have been right around
> when I tagged 0.8.0. In git, 0.9.0rc1 is meant to tag a single revision.
> In Fossil, it seems to be propagating backwards from the tagged revision
> to revisions with a previous tag.
>
> In short, Fossil doesn't seem to handle Git tags well, especially ones
> meant to tag single points in time and not propagate.
>

Here again, we are having trouble with the output generated by "git
fast-export".  For each commit, we get a single identifier that might be a
branch name, or a tag.  We never get both.  Your very first commit is called
"refs/tags/0.7.0".  I guess this means it was tagged with 0.7.0.  But there
is no branch name in the fast-export text.  So what branch is it in?  We
don't have any way of knowing, so for Fossil import, we simply assign it to
a new branch named "0.7.0".

Is 0.7.0 suppose to be a one-time tag?  Well, "git fast-export" reports that
tag on the first 41 commits of your repository.   So even if it is suppose
to be a tag on a single commit, it sure is acting like a branch.

I'm open to suggestions on how to fix this.....

Again, if you run feed the output of git-fast-export back into a fresh git
repository using git-fast-import, does it get your tagging and branch
right?  If so, please let me know, since in that cases the information is
clearly in the fast-export output text - I just haven't figured out where it
is or how to decode it.

Fortunately, these issues really shouldn't effect your repository all that
much.  Fossil is very flexible in its use of branch names and tags.  In
Fossil a "branch" is simply a commit that is tagged with a special "branch"
property.  (See
http://www.fossil-scm.org/fossil/doc/trunk/www/branching.wiki for additional
information.)  So by changing the property value, you can move check-ins
from one branch to another, freely.  In looking at repositories, like yours,
that have been transformed from Git to Fossil, clearly some of the tagging
and branch naming is not as originally intended, but in the examples I've
seen it still kind of makes sense.

Do let me know if you have any ideas on how the situation can be improved
based on what we have to work with in the output of "get fast-export".



>
> My Git repository is here:
>
> git://spielproject.info/spiel
>
> And my Fossil repository:
>
> http://dl.dropbox.com/u/147071/spiel.fossil
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.10 (GNU/Linux)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
>
> iEYEARECAAYFAkzkemgACgkQIaMjFWMehWKpLgCfZYGawVBSCNyWURDGrbfBn9MZ
> Q88AnjA6ya/FD+eE8h0sJgA8XuTegzKZ
> =66pY
> -----END PGP SIGNATURE-----
>
> _______________________________________________
> fossil-users mailing list
> [email protected]
> http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
>



-- 
D. Richard Hipp
[email protected]
_______________________________________________
fossil-users mailing list
[email protected]
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users

Reply via email to