[petsc-dev] get rid of https://bitbucket.org/petsc/petsc-dev NOW

[Patrick Sanan]

On 9/5/14 5:58 AM, Jed Brown wrote:
> Satish Balay <balay at mcs.anl.gov> writes:
>>> What's wrong with "git rev-parse --abbrev-ref HEAD" or "git symbolic-ref HEAD"?
>> balay at asterix /home/balay/petsc (maint)
>> $ git checkout origin/master
>>
>> <snip>
>>
>> balay at asterix /home/balay/petsc ((de1529e...))
>> $ git branch | sed -n '/\* /s///p'
>> (detached from origin/master)
>> balay at asterix /home/balay/petsc ((de1529e...))
>> $ git rev-parse --abbrev-ref HEAD
>> HEAD
> Yeah, less useful.
>
>> balay at asterix /home/balay/petsc ((de1529e...))
>> $ git symbolic-ref HEAD
>> fatal: ref HEAD is not a symbolic ref
> Maybe
>
>    git symbolic-ref -q HEAD || git rev-parse HEAD
>
> or
>
>    git describe --all --contains HEAD

This appears to work too:
git name-rev --name-only HEAD
(I think for detached heads it gives you master~1 or the like)
>
>> balay at asterix /home/balay/petsc ((de1529e...))
>>
>> Notice the first one has useful info - but not the others. Even
>> git-prompt gives some useful info [if not the the branch name..]
>>
>> So I pushed a change using 'git branch' and parsing its output
>> from python [instead of 'sed']
>>
>> I'm suprised 'git branch' doesnt do the most obvious thing that one
>> would expect. [default to - or have an option to list the current
>> branch - not all the branches]
> Perhaps, but it wouldn't fix the problem because there isn't always a
> "current branch".  I would rather script with plumbing commands instead
> of ones that could change output format.