Testing fastimport: changeset IDs changed unexpectedly

Greg Ward greg-hg at gerg.ca
Fri May 15 08:30:58 CDT 2009


On Fri, May 15, 2009 at 8:52 AM, Greg Ward <greg-hg at gerg.ca> wrote:
> I'll re-run my tests with and without my "overhaul backend" patch,
> using the same hg revision.  And I'll dump the changelog each time.
> Will let you know what turns up.

Found the culprit: it's mercurial_sink saving extra info about the
conversion.  If I comment out these lines:

        if self.branchnames and commit.branch:
            extra['branch'] = commit.branch
        if commit.rev:
            extra['convert_revision'] = commit.rev

from hgext/convert/hg.py, then I get my old changeset IDs back.  Your
recent changes to commit had nothing to do with it.

(But it looks like I'm doing this just in the nick of time: your
recent changes to commit break current hg-fastimport, because it uses
localrepository.rawcommit(). That will be history once I commit my
current overhaul.)

Greg



More information about the Mercurial-devel mailing list