It looks like the "Release" number is incremented and written to disk even when parsing the .spec file fails, which leads to a broken / partial update of the .spec file on disk. I think if parsing the .spec file fails, it should not be modified on disk at all.
c.f. https://bugzilla.redhat.com/show_bug.cgi?id=2068607
Retrieval of the ENVR for the changelog entry is considered unimportant enough as to handle it like a fatal error condition. Afterall, the release value is bumped properly, and a mass rebuild of packages isn't affected by missing details in a changelog entry.
Login to comment on this ticket.