For example googleapis/google-cloud-python#14738 touched all library files but does not have the extra metadata in the commit that associates the changes to all libraries -- it is therefore not captured by our bulk commit logic.

Should we have a process that says for changes like these the author should leave the libraries footer for association?

Should we change how we process releases to account for these cases? Right now we process releases on a per library basis based on source roots. Because of this a change like the one above is associated individual libraries N times as there is no metadata in the commit. Should we try to build the libraries ids based on files modified for every commit? ( this could be expensive, but maybe we could cache commits or something)