Improve handling of pushes with failures

A non-zero exit status from 'git push' doesn't necessarily mean that
nothing was pushed - we should parse the output just as for a zero
exit status. This requires adding a new option _ignore_exit_status
to git_run().

For --dry, we can actually pay attention to the return code, but need
to fix up previous logic where we returned None instead of a list of
commits, and then tried to iterate over None. For clarity, let
CalledProcessError throw through rather than special-casing a return
of None.

Inspired by a patch from Philip Withnall.

https://bugzilla.gnome.org/show_bug.cgi?id=620668

0 files changed, 0 insertions, 0 deletions