Bidi problems in KHTML

[Lars Knoll]

Hi,

the BiDi code used in khtml and Qt is almost identical (as I wrote both of 
them). It seems like I just forgot to apply one bugfix I made to Qt to khtmls 
BiDi code.

I debugged the issue, attached is a patch that should fix it. If you feel like 
debugging issues yourself, the best way is usually:

1. try to strip down the misrendered example to a minimal one, in this case 
e.g.: " بعد 2.3 لن"
2. enable BIDI_DEBUG in bidi.cpp and analyze the output.
3. Test you patch against khtmltests/i18n/bidi.html to make sure you didn't 
break anything, and add a test case there.

Hope this helps for the future.

Cheers,
Lars

On Tuesday 27 April 2004 17:32, Isam Bayazidi wrote:
> Hi all.
> The bug #56608 was reported more than a year ago, and as it seems that the
> KHTML developers do not have fixing bidi issues in thier priority lists, I
> decided to make research, and try to fix it my self:
> http://bugs.kde.org/show_bug.cgi?id=56608
>
> I am currently looking in the kdelibs/khtml/rendering/bidi.cpp to find the
> exact location of the issue, but as I am unfamiliar with the KDE code, I
> think it will take some time. I want to ask:
> - Does KHTML comply with the Unicode Annex #9 (The Bidirectional Algorithm)
> ? - what part of the code handle the cases:
> EN CS EN
> and
> AN CS AN
> ?? it seems that KHTML fails to return to the DirR after the number witht
> he CS is rendered.
>
> Can any of the KHTML developers here help me with the patch, to determine
> the exact location of the problem ?
>
> Thank you very much
-------------- next part --------------
A non-text attachment was scrubbed...
Name: bidi.diff
Type: text/x-diff
Size: 2097 bytes
Desc: not available
URL: <https://mail.kde.org/mailman/private/kfm-devel/attachments/20040429/c4e8864e/attachment.diff>