-
Emanuele Aina authored
Explicitly attempt a fast forward merge before rebasing. Currently we always try to rebase, which already doews a fast-forward merge when possible. The issue is that after that we incoditionally try to append a new changelog entry and if we're fast-forwarding we should **not** to that since the changelog will already have a version equal or newer than the one we're trying to add. By explicitly attempting a fast-forward merge and exiting early we avoid that issue. This has not been a particular issue so far since the mirroring pipeline which uses the rebasing feature does not usually trigger it when a fast-forward is possible and will create all the branches in their final state. However in some cases unrelated errors encountered during the process have left repositories in a inconsistent state and this change allows to recover cleanly from those situations. Signed-off-by:
Emanuele Aina <emanuele.aina@collabora.com>
Emanuele Aina authoredExplicitly attempt a fast forward merge before rebasing. Currently we always try to rebase, which already doews a fast-forward merge when possible. The issue is that after that we incoditionally try to append a new changelog entry and if we're fast-forwarding we should **not** to that since the changelog will already have a version equal or newer than the one we're trying to add. By explicitly attempting a fast-forward merge and exiting early we avoid that issue. This has not been a particular issue so far since the mirroring pipeline which uses the rebasing feature does not usually trigger it when a fast-forward is possible and will create all the branches in their final state. However in some cases unrelated errors encountered during the process have left repositories in a inconsistent state and this change allows to recover cleanly from those situations. Signed-off-by:
Emanuele Aina <emanuele.aina@collabora.com>