97235 - Extended Arabic letters do not join

Status: RESOLVED FIXED

(Core :: Layout: Text and Fonts, defect)

Description

[Roozbeh Pournader]

Extended Arabic letters (in the range U+0670..U+06FF) do not join properly. For
those of these who have contextual shapes in the U+FB50..U+FBFF range, the
support could be added with minor changes. This is important which languages
like Persian.

Comment 1

[Roozbeh Pournader]

Attachment: Suggested Patch

Comment 2

[Roozbeh Pournader]

Michael, would you review this?

Comment 3

[Keyser Sose]

marking NEW...this has a patch in it.

Comment 4

[Mike Kaply [:mkaply]]

Maha,

Could you look at this one?

I don't know enough about this.

Comment 5

[Maha Abou El-Rous]

I reviewed the patch, I had only one change to it and that is for the x0649 character
(alef maksoura). I changed its joining class back to eRj instead of the suggested eDj, and 
changed the check for other shapes for the character in PresentationFormB because for 
Arabic it can only be final (isolated or joined from right).
The x06CC in the Extended Arabic letters is equivelent to x0649.

Comment 6

[Maha Abou El-Rous]

Attachment: Modified patch

Comment 7

[Roozbeh Pournader]

Maha,

Please note that U+0649 (Alef Maksura) is considered "dual-joining" in the 
Unicode Standard. You can see that yourself at: 
<http://www.unicode.org/Public/UNIDATA/ArabicShaping.txt> where it is listed as:

   0649; ALEF MAKSURA; D; YEH
                      ^^^
Please note that I, as a Persian speaker, am not interested in this character, 
since it is not used in Persian. The Yeh character used in Persian is U+06CC.

But I am interested in proper Unicode support. U+0649 is used in the "Initial" 
and "Medial" forms also, in some of the countries formerly in USSR. If you 
think Unicode is wrong, ask them to fix it; if it is not wrong, please let 
Mozilla have bugless Unicode support.

Note: I don't have any problem with any patch, if that solves the Persian 
problem. So I think I agree with Maha's patch. I can open a new bug for U+0649 
Arabic Letter Alef Maksura.

Comment 8

[Roozbeh Pournader]

Oops, found a typo in the patches. '0x06D3' was listed as '0x0D3'. Will attach 
a new patch based on Maha's patch.

Comment 9

[Roozbeh Pournader]

Attachment: Final patch, fixing a typo

Comment 10

[Maha Abou El-Rous]

Roozbeh, appreciate the info. x0649 was defined as R in Unicode version 3.0, I'm 
contacting Unicode org. to check why it got changed later. I agree we should seperate this 
issue from this bug. Opened a new bug for it # 99256

Mike, Would that be considered a review, should I mark it as such? 

Comment 11

[Mike Kaply [:mkaply]]

Yep, that's a review and I'll consider it such. I'll look for a super 
refviewer. We should open a separate bug on the other issue.

Comment 12

[Roozbeh Pournader]

Michael, any news from a super reviewer?

Comment 13

[Roozbeh Pournader]

Michael, would you please give me an "r="? I will find a super reviewer myself.

Comment 14

[Mike Kaply [:mkaply]]

Actually, I had r=, I'll put it in the bug, and I had asked a super reviewer. 
I'll ask another one.

Comment 15

[Roozbeh Pournader]

Attachment: Cleaned up patch, against 0.9.5

Comment 16

[Roozbeh Pournader]

The third patch (id 48855) could not be applied on Linux, because the filename
cases were changed. Created a new attachment with no change but the filenames
and line numbers that it can now be applied on Linux systems. It is made against
0.9.5 and is tested to work.

Comment 17

[Mike Kaply [:mkaply]]

Comment on attachment 55522 [details] [diff] [review]
Cleaned up patch, against 0.9.5

r=mkaply

Comment 18

[Christopher Blizzard (:blizzard)]

Comment on attachment 55522 [details] [diff] [review]
Cleaned up patch, against 0.9.5

sr=blizzard

Comment 19

[Mike Kaply [:mkaply]]

Fix finally checked in