A peculiar GoB bug on this CL made me wonder: what should bot_update do when applying a patch in two special cases:

1. If patch is empty - ie no files modified in it (example)

2. If patch has already been committed, but CQ is not aware of it (like here) or say tryjob is pending for too long time.

In both cases, the diff will be empty after rebase. Such patches should be rejected to avoid developers and tooling confusion by failing with PATCH_FAILURE exit code and a clear message.
 

Consider bot_update for Gerrit, which has just rebased patch against HEAD in the following two cases:

1. If patch is empty - ie no files modified in it (example).
2. If patch has already been committed, but CQ is not aware of it (like here) or say tryjob is pending for too long time.

In both cases, the diff will be empty after rebase. Such patches should be rejected to avoid developers and tooling confusion by failing with PATCH_FAILURE exit code and a clear message.