[Bug 4514] New: Significant performance regression when running hg diff -g

[mercurial-bugs at selenic.com]

http://bz.selenic.com/show_bug.cgi?id=4514

          Priority: normal
            Bug ID: 4514
                CC: mercurial-devel at selenic.com
          Assignee: bugzilla at selenic.com
           Summary: Significant performance regression when running hg
                    diff -g
          Severity: bug
    Classification: Unclassified
                OS: Mac OS
          Reporter: zenbot at gmail.com
          Hardware: PC
            Status: UNCONFIRMED
           Version: 3.3-rc
         Component: Mercurial
           Product: Mercurial

Hi there,

We've seen a significant performance regression running hg di with git extended
diff format on the 3.3 release candidate. On 3.2.4 we see:

$ time hg di -r ce7ddc8ad8bc:default spirit -g >/dev/null
        2.64 real         2.50 user         0.13 sys

On 3.3-rc we see:

$ time hg di -r ce7ddc8ad8bc:default spirit -g >/dev/null
      216.96 real       212.72 user         4.14 sys

Bisect identifies changeset c48924787eaa as the culprit.

I wasn't able to reproduce this with a simple test repository. The repo we're
seeing this performance on has approximately 70,000 commits, a ton of branches,
and uses largefiles. In the example above there's about 60,000 commits between
the two revisions being diff'd and "spirit" is a directory.

Let me know if I can assist further.

Cheers,

Andrew

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