Bug 34247 - [Gtk] Verify identical behavior of ICU and GLib unicode backends
Summary: [Gtk] Verify identical behavior of ICU and GLib unicode backends
Status: RESOLVED INVALID
Alias: None
Product: WebKit
Classification: Unclassified
Component: Platform (show other bugs)
Version: 528+ (Nightly build)
Hardware: PC Linux
: P2 Normal
Assignee: Nobody
URL:
Keywords:
Depends on: 48056 48227
Blocks:
  Show dependency treegraph
 
Reported: 2010-01-28 01:16 PST by Dominik Röttsches (drott)
Modified: 2013-02-13 04:14 PST (History)
5 users (show)

See Also:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Dominik Röttsches (drott) 2010-01-28 01:16:14 PST
There are open FIXME's regarding the behavior of the GLib unicode backend compared to to the ICU one. 

Test cases need to be created to ensure identical behavior of both for cases mentioned in TextBreakIteratorGtk.cpp, such as:

1) FIXME: assumes no surrogate pairs

2) FIXME: UBRK_WORD case: Single multibyte characters (i.e. white space around them), such as the euro symbol â¬,
are not marked as word_start & word_end as opposed to the way ICU does it. This leads to - for example - different word selection behaviour when right clicking.

3) FIXME: This needs closer inspection to achieve behaviour identical to the ICU version.

When this has been done, we should be able to remove the configure-time stability warning for the GLib unicode backend that's currently autotools/webkit.m4.
Comment 1 Fridrich Strba 2010-03-03 14:59:37 PST
is the WTF_USE_GLIB_ICU_UNICODE_HYBRID define still needed? I see two places in ThreadGlobalData.h where one still mentions it, but it seems to be cleared from ThreadGlobalData.cpp though.
Comment 2 Dominik Röttsches (drott) 2010-03-04 01:23:13 PST
(In reply to comment #1)
> is the WTF_USE_GLIB_ICU_UNICODE_HYBRID define still needed? I see two places in
> ThreadGlobalData.h where one still mentions it, but it seems to be cleared from
> ThreadGlobalData.cpp though.

You're right, this is orphaned and should go away. I created bug #35726 and attached a patch for review. Thanks for the reminder.
Comment 3 Dominik Röttsches (drott) 2013-02-13 04:14:41 PST
GLib backend is gone per bug 109627, closing.