Details
-
Bug
-
Resolution: Done
-
P1: Critical
-
4.5.0
-
None
-
106148704ee1e73c7b71c4a1a2dcb64a1ac52727
Description
Steps to reproduce / test case
Load the text file:
http://sites.google.com/site/jamadagni/files/indic-two-part-vowel-signs.txt (UTF-8)
in some Qt application. The actual rendering (based on KDE 4.2 / Qt 4.5 on reporters Kubuntu Jaunty system) is
http://sites.google.com/site/jamadagni/files/indic-two-part-vowel-signs.png
where the desired rendering is as on the right side of the image, rendered by Pango for GTK. Pango only faults with Sinhala EE and OO which I will report as a bug for Pango. You guys get busy fixing Indic rendering in Qt!
Background: Unicode provides decompositions for all these Indic vowel signs. A decomposition of a vowel sign and the separately encoded vowel sign should be rendered the same. This was a small study of the current support of such decomposed vowel signs in Indic scripts. Pango wins much to my chagrin as a Qt/KDE supporter. Please fix this for 4.6 if possible.
There some Indic vowel signs which in fact can be input as three components. They are:
Kannada: 0CCB Vowel Sign OO = 0CCA Vowel Sign O + 0CD5 Length Mark = 0CC6 Vowel Sign E + 0CC2 Vowel Sign UU + 0CD5 Length Mark (because 0CCA Vowel Sign O = 0CC6 Vowel Sign E + 0CC2 Vowel Sign UU)
Sinhala: 0DDD Vowel Sign OO = 0DDC Vowel Sign O + 0DCA Virama = 0DD9 Vowel Sign E + 0DCF Vowel Sign AA + 0DCA Virama (because 0DDC Vowel Sign O = 0DD9 Vowel Sign E + 0DCF Vowel Sign AA)
Inquiry on the indic@unicode.org mailing list reveals that these three-component decompositions are indeed to be rendered equivalently as their NFD normalization form is the same.
So along with including support for the first-level decompositions (consisting of two codepoints) please also provide the same for the second-level decompositions (consisting of three codepoints).