[Bug 6538] New: Follow up to issue 6528, parent ordering

mercurial-bugs at mercurial-scm.org mercurial-bugs at mercurial-scm.org
Mon Jul 12 15:04:38 UTC 2021


https://bz.mercurial-scm.org/show_bug.cgi?id=6538

            Bug ID: 6538
           Summary: Follow up to issue 6528, parent ordering
           Product: Mercurial
           Version: default branch
          Hardware: PC
                OS: Linux
            Status: UNCONFIRMED
          Severity: bug
          Priority: wish
         Component: Mercurial
          Assignee: bugzilla at mercurial-scm.org
          Reporter: pierre-yves.david at ens-lyon.org
                CC: mercurial-devel at mercurial-scm.org
    Python Version: ---

issue6528 was introduced by 49fd21f32695 :

  changeset:   49092:49fd21f32695
  user:        Joerg Sonnenberger <joerg at bec.de>
  date:        Wed Mar 10 18:09:21 2021 +0100
  summary:     revlog: guarantee that p1 != null if a non-null parent exists

A changeset that was eventually backed out last week as 411dc27fd9fd

To sumup 49fd21f32695, swapping p1 and p2 is p1 is nullid destroy the implicit
"this file-revision has copy information" marker and corrupt repository.

However Joerg Sonnenberger have been warning use about other potential bug that
backing out this changeset introduce.

such bug would be related to bad order of changelog-revision, however I don't
have more details about them.

I think we needs:
- clearer information from Joerg about these changelog related bug (or someone
else to dig them out),

- a new changeset to prevent parent swapping on filelog, but allowing it on
changelog.

- a script to fully scrub a repository to detect and swapped filelog
information.

-- 
You are receiving this mail because:
You are on the CC list for the bug.


More information about the Mercurial-devel mailing list