Re: [PATCH 2/3] remote-hg: allow invalid bookmarks in a few edge cases

[Antoine Pelisse <apelisse@gmail.com>]

Hi Max,

Thank you for working on this.
I believe it would be fair that you forget about patch 1/3 as you fix
it in this patch (2/3).
Also, I think it would be best NOT to integrate a patch (mine) that
breaks a test, as it
would make bisect harder to use.

Thanks,
Antoine

On Wed, Mar 19, 2014 at 1:33 PM, Max Horn <max@quendi.de> wrote:
> Fix the previous commit to workaround issues with edge cases: Specifically,
> remote-hg inserts a fake 'master' branch, unless the cloned hg repository
> already contains a 'master' bookmark. If that 'master' bookmark happens
> to reference the 'null' commit, the preceding fix ignores it. This
> would leave us in an inconsistent state. Avoid this by NOT ignoring
> null bookmarks named 'master' or 'default' under suitable circumstances.
>
> Signed-off-by: Max Horn <max@quendi.de>
> ---
>  contrib/remote-helpers/git-remote-hg | 7 +++++--
>  1 file changed, 5 insertions(+), 2 deletions(-)
>
> diff --git a/contrib/remote-helpers/git-remote-hg b/contrib/remote-helpers/git-remote-hg
> index 12d850e..49b2c2e 100755
> --- a/contrib/remote-helpers/git-remote-hg
> +++ b/contrib/remote-helpers/git-remote-hg
> @@ -626,8 +626,11 @@ def do_list(parser):
>      repo = parser.repo
>      for bmark, node in bookmarks.listbookmarks(repo).iteritems():
>          if node == '0000000000000000000000000000000000000000':
> -            warn("Ignoring invalid bookmark '%s'", bmark)
> -            continue
> +            if fake_bmark == 'default' and bmark == 'master':
> +                pass
> +            else:
> +                warn("Ignoring invalid bookmark '%s'", bmark)
> +                continue
>          bmarks[bmark] = repo[node]
>
>      cur = repo.dirstate.branch()
> --
> 1.9.0.7.ga299b13
>