for hg and git we're re-creating the E_HEAD_IN_BRANCH_TO_REVID after each slurp that found new changesets. These indexes however aren't being re-created. As a result revisions that are silently merged in (i.e. without conflicts) are missed by this cond...

Classification: Factually incorrect

Created by Michael Heemskerk from Crucible defect on WaybackSpec.java line 117 https://extranet.atlassian.com/crucible/cru/CR-FE-4154#c36010 :

for hg and git we're re-creating the E_HEAD_IN_BRANCH_TO_REVID after each slurp that found new changesets. These indexes however aren't being re-created. As a result revisions that are silently merged in (i.e. without conflicts) are missed by this condition.

This query is only being used for determining whether a file is empty. When getting the file contents a different query is being used, which is wrong I think...

Marked as defect because I need to track down usages of E_UNMOD_ON_BRANCH_TO_REVID and possibly refactor them or re-generate the E_UNMOD_ON_BRANCH_.. as well