GNOME Bugzilla – Bug 302952
The placement of a diacritic marks for an arabic ligature is not correct
Last modified: 2007-12-03 14:13:44 UTC
Please describe the problem: The placement of an arabic diacritic marks is not correctly done when a ligature is formed. This happens especially the first character does not have mark associate to it. The font is an opentype font, table created using VOLT. For example, the sequence is as follows: ain kasra yeh noon damma . The yeh and noon is correctly converted to a new ligature, but the placement of damma is not right. It is placed above the first anchor of the ligature (above yeh instead of noon). Steps to reproduce: 1. 2. 3. Actual results: Expected results: Does this happen every time? No, at some place it placed correctly at the second anchor of the ligature Other information: E-mail me for a sample file and font.
Created attachment 46003 [details] Picture comparing the result for windows and gnome The left side the the picture taken from notepad under windows, and the right side is the picture for gedit under Ubuntu 5.04 (gnome 2.10, pango 1.8.1).
Does this happen with other fonts too? Can you attach the font?
Have specifics on the font is really likely necessary for progress here. Otherwise, all that is possible to do is to read the source code and hope that it is obvious.
At the moment, I can't really give you the font that I'm working on, maybe later on. However, if you really insist on having a sample font, you can use the one from Microsoft VOLT community. Anyone can joind the community for free, but the sample font is restricted (thus I can't give upload that font either) I can give you the sample data if you need it. Using the font from VOLT, you can see that the placement of diacritics marks is not correct as compared to windows behaviour, similar to the screenshot that I've provided. Hope that will help to tackle this problem.
Thanks for the info (or lack there of). I can reproduce the problem slightly differently with the font from VOLT, but since there's no freely redistributable font showing this problem, I think this bug can be resolved as INCOMPLETE.
What do you mean by resolved as INCOMPLETE ? Personally, I really think there is a bug somewhere, and I really would like it to be resolved. And as for the free font, I'm planning to release one soon, but at the moment it is incomplete. Maybe in 2 weeks time.
By resolving to "INCOMPLETE" I mean there's something missing that prevents one from trying to resolve this bug as "FIXED". It's not even obvious that there is a bug in Pango. It may be a deviation from the OpenType spec in Uniscribe, like bug 170180 is. If you really like this to be fixed, you can provide a minimal font that exposes the problem, and I will be more than happy to try to fix it. I already spent a couple hours, but gave up.
The same bug happens with Arabeyes fonts. I noticed that in the first ligature in the line, whatever the ligature is, anchor points are ingored and diacritic marks are put over the first component of the ligature. I tested it with 3 fonts and I got the same result.
Created attachment 91940 [details] Screenshot in this screenshot, each 1st ligature in the line isn't behaving as expected, while the second ligature which is essentially the same as the 1st one but is working properly.
Created attachment 91941 [details] Sample font The font used in the screenshot above, I've onother 3 fonts with the same behaviour.
*** This bug has been marked as a duplicate of 437633 ***
Ok, fixed the problem with first ligature in line. The rest of the issues will be handled in bug 437633. 2007-08-29 Behdad Esfahbod <behdad@gnome.org> Bug 302952 – The placement of a diacritic marks for an arabic ligature is not correct * pango/opentype/harfbuzz-buffer.c (hb_buffer_allocate_ligid): Don't use zero as allocated ligature id. Zero means no ligature id.