1275474 - nsIUnicodeNormalizer should use ICU library if turned on

Status: RESOLVED INCOMPLETE

(Core :: Internationalization, defect)

Description

[Makoto Kato [:m_kato]]

For reducing table...

Comment 1

[Makoto Kato [:m_kato]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=59d7f9522168811f1b8671ef30a6a379aab0abed

Comment 2

[Makoto Kato [:m_kato]]

Attachment: MozReview Request: Bug 1275474 - Part 1. nsIUnicodeNormalizer should use ICU library if turned on. r?jfkthame

By bug 1261900, Compress and DecomposeNonRecursively are no logner used when ICU is turned on.  So it shouldn't be used with ICU.

Review commit: https://reviewboard.mozilla.org/r/55014/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/55014/

Comment 3

[Makoto Kato [:m_kato]]

Attachment: MozReview Request: Bug 1275474 - Part 2. Add nsIUnicodeNormalizer unit test. r?jfkthame

Review commit: https://reviewboard.mozilla.org/r/55016/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/55016/

Comment 4

[Jonathan Kew [:jfkthame]]

IIRC, currently nsIUnicodeNormalizer implements an old version (3.2?) of Unicode, because of legacy IDN considerations. (I wanted to update it at one point -- there's probably still an open bug -- but the IDN dependency was a problem.)

If we do this, updating the behavior of nsIUnicodeNormalizer when ICU is in use (which also means we're using ICU for IDN processing), then we'll be exposing different normalization behavior on Android vs other platforms..... Does that matter?

Comment 5

[Jonathan Kew [:jfkthame]]

(In reply to Makoto Kato [:m_kato] from comment #2)
> By bug 1261900, Compress and DecomposeNonRecursively are no logner used when
> ICU is turned on.  So it shouldn't be used with ICU.

Bug-number typo: just FTR, the relevant bug there is 1161900.

Comment 6

[Makoto Kato [:m_kato]]

(In reply to Jonathan Kew (:jfkthame) from comment #4)
> IIRC, currently nsIUnicodeNormalizer implements an old version (3.2?) of
> Unicode, because of legacy IDN considerations. (I wanted to update it at one
> point -- there's probably still an open bug -- but the IDN dependency was a
> problem.)

Ah OK.

> If we do this, updating the behavior of nsIUnicodeNormalizer when ICU is in
> use (which also means we're using ICU for IDN processing), then we'll be
> exposing different normalization behavior on Android vs other platforms.....
> Does that matter?

I see.  Before it, I should integrate ICU (Bug 1262102) at first.  Until landing it, I should cancel this.

Comment 7

[Makoto Kato [:m_kato]]

Also, this XPCOM uses some addons and plugin code (NPNVdocumentOrigin)...

Comment 8

[Makoto Kato [:m_kato]]

I removed nsIUnicodeNormalizer