Tom Lane wrote:
> Heikki Linnakangas <heikki.linnakangas@enterprisedb.com> writes:
>
>> Cédric Villemain wrote:
>>
>>> 2010/4/30 Stefan Kaltenbrunner <stefan@kaltenbrunner.cc>:
>>>
>>>> I don't think the git repo was ever considered working for the backbranches
>>>> at all...
>>>>
>>> Really ?!
>>> Then we have to remove the backbranches from the git.
>>> http://wiki.postgresql.org/wiki/Working_with_Git#Using_Back_Branches
>>>
>
>
>> Yeah, or fix them.
>>
>
> This thread isn't exactly leaving me with a warm fuzzy feeling about
> moving the master repository to git. *Why* is the mirror broken, and
> what assurances do we have that the problem won't recur in the proposed
> transition?
>
>
>
AFAICT it is broken because the particular tool that is used, which is
the only one that supports an incremental mode, is a bit broken. I am
told that the non-incremental tools are more robust.
That said, this is more than a little annoying. It means, for example,
that I can't test out a Git mode for the buildfarm client on all the
back branches.
If any Ruby hacker feels like fixing it please speak up. The reported
source of the software seems to have gone away. I can let you have my
copy, which reliably reproduces the error, so we have a good failure
test case.
cheers
andrew