554609 - fts3 tokenizer should use bi-gram for Thai, Lao and Khmer

Status: RESOLVED FIXED

(MailNews Core :: Database, defect)

Description

[Makoto Kato [:m_kato]]

Thai, Lao and Khmer is complex word-break rule.  These language should use bi-gram like CJK.

U+0E00-U+0E7F ... Thai
U+0E80-U+0EFF ... Lao
U+1780-U+17FF ... Khmer

Comment 1

[Makoto Kato [:m_kato]]

Attachment: patch

Comment 2

[Andrew Sutherland [:asuth] (he/him)]

Comment on attachment 435554 [details] [diff] [review]
patch

Looks good.  This can land now but let's not mark a schema change yet.  Testers will need to delete their database manually.

Comment 3

[Ludovic Hirlimann [:Usul]]

(In reply to comment #2)
> (From update of attachment 435554 [details] [diff] [review])
> Looks good.  This can land now but let's not mark a schema change yet.  Testers
> will need to delete their database manually.

Andrew which bug should make the schema change ?

Comment 4

[Andrew Sutherland [:asuth] (he/him)]

(In reply to comment #3)
> Andrew which bug should make the schema change ?

A new bug should be filed if explicit tracking is desired.  Since we already have a schema change in the build since beta 1, anyone upgrading from beta 1 to beta 2 when it ships will already have this taken care of for them.  This leaves nightly users, and I'm expecting another schema change to come down the nightly pipe not too long from now, so I wasn't going to worry about it.  Given my understanding of the limited numbers of 3.1 nightly users, it didn't strike me that more than a handful of users would be likely to be affected.

Comment 5

[Makoto Kato [:m_kato]]

landed
http://hg.mozilla.org/comm-central/rev/f3ae26af4fb8