Per http://www.sqlite.org/fts3.html, the doclist encoding for terms encountered in the 0th column is specialized. This would save us 2 bytes per distinct term per document in the body. It would also cost us 2 bytes per term per document in the subject, but since the subject should generally be much shorter than the body, the choice is clear.
This seems to make test_msg_search.js fail (the non-failing tests might by commented out). Any idea as to why this may be failing will be highly appreciated :-).
Assignee: bugmail → jonathan.protzenko
Attachment #551971 - Flags: feedback?(bugmail)
Ok, found what I missed. I fixed the comment if we ever want to swap the two again ;-)
Comment on attachment 552003 [details] [diff] [review] Patch v2 Ah yeah, the ranker function was a more recent addition. Looks great, thanks for taking care of this!
Attachment #552003 - Flags: review?(bugmail) → review+
I suspect you are already aware, but just to make sure, we cannot land this unless we also land a patch that bumps the schema version, so do keep it in your queue. I think it's okay if we bump the schema multiple times on comm-central as long as it doesn't get shifted to comm-aurora in between, so let's wait to do that until just after we make the next cut from comm-central to comm-aurora which should be happening soon.
Status: ASSIGNED → RESOLVED
Closed: 8 years ago
Resolution: --- → FIXED
Target Milestone: --- → Thunderbird 10.0
You need to log in before you can comment on or make changes to this bug.