GNOME Bugzilla – Bug 380094
Fix split matra rendering bug for Kannada language
Last modified: 2008-06-23 21:30:15 UTC
Please describe the problem: Pango rendering splits the glyphs U+0C8E (ಎ) and U+0C82 (anusvaara) when they appear in sequence. Steps to reproduce: 1. Type the word ಎಂಬ ("eMba" with SCIM-m17n kn-itrans selection). 2. Paste the same on Kate or Kwrite. Actual results: You'll notice the difference in rendering of the split maatra. Expected results: Pango should render the Anusvaara (U+0C82) beside U+0C8E (ಎ)) without dotted circles. Does this happen every time? Yes. Other information: http://sampada.net/pango-kannada-rendering-bug.zip contains two images - one taken on QT based application that shows correct rendering and the other showing the bad rendering on Pango (the files are appropriately named). Take a note of just the first two letters on those.
I'm confirming the bug mentioned as above. I copied the above word "ಎಂಬ" into gedit, gaim and the rendering remained the same.
(In reply to comment #1) > I'm confirming the bug mentioned as above. I copied the above word "ಎಂಬ" > into gedit, gaim and the rendering remained the same. > i confirm the same.
Created attachment 79710 [details] [review] The patch solves the problem , mentioned in the bug patch taken against pango-1.14.9
Was previously defined as an unknown character (_xx) in the indic-ot-class-table.c file.
Did any one get a chance to apply this patch?
The patch released on 2007-01-08 by Mr. Ramakrishna works pretty well with the new development release too Pango Version 1.18.1.
Same bug is also reported at http://bugzilla.gnome.org/show_bug.cgi?id=481205 This is an important bug with a long pending trivial patch verified by so many reporters and discussed in so many places. Please commit it and close it.
The patch has been submitted about 8 months back and people have certified that its working fine, but still, there is hardly any answer by the upstream!!? Can we know what is the roadblock for this patch to get comitted/accepted!!? Some thing more should be done from our side!!?
Behdad, this patch is very simple and well tested as well now. I think it is worth committing now.
The patch is simple, well tested, but no one even tried explaining *why* it's the correct patch. In fact, in this simple case my untrained Indic eyes see that it's not the correct patch. Attaching correct one.
Correct patch. The thing is, U+0C8D is Reserved, but the Indic table was marked as if U+0C8E is such. Swapping the classes. Index: indic-ot-class-tables.c =================================================================== --- indic-ot-class-tables.c (revision 2657) +++ indic-ot-class-tables.c (working copy) @@ -196,7 +196,7 @@ static const IndicOTCharClass teluCharCl */ static const IndicOTCharClass kndaCharClasses[] = { - _xx, _xx, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _iv, /* 0C80 - 0C8F */ + _xx, _xx, _mp, _mp, _xx, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _iv, _xx, _iv, _iv, /* 0C80 - 0C8F */ _iv, _xx, _iv, _iv, _iv, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, /* 0C90 - 0C9F */ _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _bb, /* 0CA0 - 0CAF */ _rb, _ct, _bb, _bb, _xx, _bb, _bb, _bb, _bb, _bb, _xx, _xx, _xx, _xx, _dr, _da, /* 0CB0 - 0CBF */
2008-06-23 Behdad Esfahbod <behdad@gnome.org> Bug 380094 – Fix split matra rendering bug for Kannada language * modules/indic/indic-ot-class-tables.c: Fix Indic class for U+0C8D and U+0C8E.