Status: New
Owner: ----
Labels: Type-Defect Priority-Medium

New issue 2255 by Issue with the display of a diff in presence of a parent diff

* NOTE: Do not post confidential information in this bug report. *
*       If you need immediate support, please contact            *
*                             *

What version are you running?

RB 1.5.1 and RB 1.5.2.  Issue also present in RB 1.6RC2.

What's the URL of the page containing the problem?


What steps will reproduce the problem?

1. Create mercurial repository to which RB has access to and a local clone.
2. Create an initial changeset as indicated by this output:
plato:sandbox ballarin$ hg up 91c24aa2a306
1 files updated, 0 files merged, 0 files removed, 0 files unresolved
plato:sandbox ballarin$ cat mq-test.txt
File for investigating an issue of the mercurial reviewboard extension.

Posting a request may result in a patch being displayed incorrectly if
an ancestor changeset has changed in the meantime.  This happens
regularly with patch queues.

First patch

Some space for preventing
a conflict between the
first and the second patch.

Second patch
3. Add a changeset that changes "First patch" to "First patch 1"
4. Add another changeset that changes "Second patch" to "Second patch 2"
5. Push only the initial changeset to the server. The graph is now like this:

91c24aa2a306 (on server)
| parent diff
cfbda047a23c (local)
| diff
e86b4f76fcd0 (local)

6. Upload patch e86b4f76fcd0 with cfbda047a23c as parent diff.

What is the expected output? What do you see instead?

In the attached screen shot, the left column is identified as revision 91c24aa2a306, while it is actually revision cfbda047a23c. (The line "First patch 1" is in cfbda047a23c, while 91c24aa2a306 actually contains "First patch".)

This is for mercurial. To exclude problems with the reviewboard extension of mercurial, I uploaded the diffs manually.

The observed behaviour is an issue when using mercurial patch queues for working simultaneously on several patches. Then both diff and parent diff can change, and when clicking through the diffs in RB we spotted that the content of the left side changed while this was always identified as the same revision.

What operating system are you using? What browser?

not relevant

Please provide any additional information below.

Fixes for this issue are proposed at Christian Hammond reported performance issues with the SQLMutation in revision 2 when migrating medium-sized deployments. Revisions 3 and 4 don't use SQLMutations and should not exhibit the problem.
Unit tests for this are proposed at

        diff.png  194 KB

You received this message because you are subscribed to the Google Groups 
"reviewboard-issues" group.
To post to this group, send email to
To unsubscribe from this group, send email to
For more options, visit this group at

Reply via email to