Merge lp:~danilo/loggerhead/bug-718566 into lp:loggerhead
Proposed by
Данило Шеган
Status: | Merged |
---|---|
Merged at revision: | 455 |
Proposed branch: | lp:~danilo/loggerhead/bug-718566 |
Merge into: | lp:loggerhead |
Diff against target: |
109 lines (+4/-38) 3 files modified
loggerhead/apps/branch.py (+2/-5) loggerhead/changecache.py (+0/-18) loggerhead/history.py (+2/-15) |
To merge this branch: | bzr merge lp:~danilo/loggerhead/bug-718566 |
Related bugs: |
Reviewer | Review Type | Date Requested | Status |
---|---|---|---|
John A Meinel | Approve | ||
Review via email: mp+71849@code.launchpad.net |
Description of the change
= Bug 718566 =
Remove FileChangeCache because it's not needed anymore (as suggested by John in bug 718566, comment #2).
All tests pass, and ./serve-branches browsing to a revision keeps working.
I'd welcome any ideas on how to best QA this.
To post a comment you must log in.
The only reason we needed the file changes cache was because formats pre-2.0 were much slower about finding what files have changed (as mentioned in the comment).
The places this shows up is stuff like when you expand a revision on "Changes" and it shows you what files were changed in that revision.
There shouldn't be an effect on correctness. The big question is one of performance. So the best bet is to run loggerhead on a large project (launchpad could be large enough). Then expand some revisions, to see it churn on showing the file list. Then restart loggerhead, and time the same queries.
With the cache, it should certainly be faster. But without a cache if it is *fast enough* then we don't need the cache.
Overall, I approve of this change, given that I suggested it. But it would be good to get some hard performance data before we roll this out on Launchpad.