Re: [PATCH v3 1/3] t3201: test multiple branch filter combinations

> A few comments:
>
> I didn't examine it too closely, so this may be a silly question, but
> is there a reason to start from scratch (by deleting all the branches)
> rather than simply using or extending the existing branches like the
> other tests do?
>
> If it really does make sense to start from scratch (ignoring the
> existing branches), then an alternative would be to create a new
> repository and run the tests in that repository instead. Whether or
> not doing so makes sense in this case is a judgment call. For
> instance:
>
>     test_create_repo features
>     (
>         cd features
>         ...setup stuff...
>     )

> It's a bit concerning to see output from porcelain git-branch being
> fed to 'grep' and 'xargs'. More typically, you would instead rely upon
> the (stable) output of a plumbing command. For instance:
>
>     git for-each-ref --format="%(refname:short)" refs/heads/ | ...
>
> In new test code, normally avoid having a Git command upstream of a
> pipe since its exit code will be lost. Thus, you might instead write:
>
>     git for-each-ref ... >heads &&
>     grep -v master heads | xargs git branch -D &&