WebKit Bugzilla
New
Browse
Search+
Log In
×
Sign in with GitHub
or
Remember my login
Create Account
·
Forgot Password
Forgotten password account recovery
RESOLVED FIXED
5597
REGRESSION (412.2-416.12): [[xxx]] appears on page instead of JS correctly parsing translated variables
https://bugs.webkit.org/show_bug.cgi?id=5597
Summary
REGRESSION (412.2-416.12): [[xxx]] appears on page instead of JS correctly pa...
Geoffrey Garen
Reported
2005-11-02 11:12:52 PST
This bug is also in Radar as <
rdar://4323773
> The latest OS update causes a regression bug in javascript. This is a bug that I noticed recently in the webkit (just started to do my own builds)... and should be (IMHO) be considered a show-stopper. The broken javascript functionality is best seen if you try to view a TiddlyWiki web page. You can go to the TiddlyWiki website to view <
http://www.tiddlywiki.com/
> I have not yet tried to pinpoint where the broken methods are.... but you will notice that there are a lot of [[foobar]] phrases on the site... where the browser should instead be parsing out translated variables. This regression can be seen if compared to the 10.4.2 OS release of the webkit.
Attachments
Testcase?
(323 bytes, text/html)
2005-11-02 14:28 PST
,
mitz
no flags
Details
Testcase!
(410 bytes, text/html)
2005-11-03 08:00 PST
,
mitz
no flags
Details
Show Obsolete
(1)
View All
Add attachment
proposed patch, testcase, etc.
Darin Adler
Comment 1
2005-11-02 13:20:59 PST
Bug 5571
contains an earlier report of this same thing (on a page that shows an even worse symptom on TOT).
mitz
Comment 2
2005-11-02 14:28:47 PST
Created
attachment 4567
[details]
Testcase? This tests for the regression in 416.12 that causes the page to break. That regression does not exist in TOT, where the page breaks for another reason (a separate RegEx regression in TOT).
Darin Adler
Comment 3
2005-11-02 14:33:34 PST
So the bug here is that if a regular expression character class has any characters > 255 in it, it won't match anything. Instead the intent was to ignore characters > 255, but it seems that instead we do something wrong with the entire character class. Should be straightforward to fix, but not on TOT, since this is only for 10.4.3 (so we might not want to use bugzilla to track this 10.4.3-only issue any more).
Darin Adler
Comment 4
2005-11-02 14:52:19 PST
What we do in 10.4.3 with characters > 255 is reject the entire regular expression. I guess that's what we need to fix. We can probably instead turn anything > 255 into a 255.
Darin Adler
Comment 5
2005-11-02 20:20:30 PST
With my patch for
bug 5602
in place, the reduction here works fine as does the test case attached to that bug. But
http://tiddlywiki.com
still doesn't work.
mitz
Comment 6
2005-11-03 07:49:55 PST
(In reply to
comment #5
)
> With my patch for
bug 5602
in place, the reduction here works fine as does the test case attached to
that
> bug. But
http://tiddlywiki.com
still doesn't work.
Looks like TOT has a problem with characters >255 in RegExp as well (but one that this bug's testcase doesn't detect), since with the current patch for
bug 5602
, if you replace all such characters in regex patterns in the tiddlywiki source with \u00ff, it works.
mitz
Comment 7
2005-11-03 08:00:43 PST
Created
attachment 4580
[details]
Testcase! This slightly more elaborate testcase fails on TOT as well.
Geoffrey Garen
Comment 8
2005-11-06 12:43:53 PST
The fix for 5571 fixes this.
Note
You need to
log in
before you can comment on or make changes to this bug.
Top of Page
Format For Printing
XML
Clone This Bug