Hi,
I have shared the relevent files in this Google document folder (url below)
https://docs.google.com/leaf?id=0B9VHTVzubonwOWU4MzdjNDEtOWMyNi00Nzk4LTk1YTItZmMzYWIzNDcyMjU3&hl=en
Though there are several files in that folder, you need to work with
only few, hence do not worry.
1. First download the test file kshproblem.txt for Telugu or use the
following step to convert it and create your test file.
2. Transliterate this to your language using Firefox(Chrome did not
work) at the site. http://girgit.chitthajagat.in/
with the input as
http://font-render.blogspot.com/2010/07/ksh-akhand-rendering-test-page.html
and selected desired language. After conversion, copy the converted
text to a text file kshproblem-<lang code>.txt
3. Open the language test file with gedit and note if there are any
problems in the language under test
4. Download debian binaries
libpango1.0-0_1.28.0-0ubuntu3_i386.deb5. Install by double clicking on them
libpango1.0-common_1.28.0-0ubuntu3_all.deb
I have given additional files in the share folder, in case there are
more dependencies than what I listed above. (It is my first attempt to
build debian packages).
6. Check whether packages are updated (you will see 1.28.0-0ubuntu3 when
you give the command in fully maximized terminal
$ dpkg -l "*pango*" )
ii libpango1.0-0 1.28.0-0ubuntu3 Layout
and rendering of internationalized text
ii libpango1.0-common 1.28.0-0ubuntu3
Modules and configuration files for the Pango
7. open kshproblem.txt or the transliterated version (step 2) with
gedit and check any problem with akhands in your language
8. Use Firefox or any software which uses Gnome for rendering of text
in language under test by visiting wikipedia or other unicode
websites. (Firefox is broken for Telugu rendering by repeating words
with this fix). Check with Chrome if you have problem with Firefox.
9. Please share your testing results along with the language font
(Lohit or others) you used by replying to this thread or posting it as
a comment on the Telugu bug
If you are interested in building from source, please use the file
pango1.0-1.28.1.tar.bz2 or apply the patch
https://bugzilla.gnome.org/attachment.cgi?id=135652 at Telugu bug to
indic-fc.c
The known issues of the fix: Character positioning is happening over
the character cluster apart from character cluster edges in gedit. If
you have any tips to fix this, please share.
For more on font rendering status: http://behdad.org/text/
Thanks in advance again for your support
Regards
Arjun
2010/7/4 arjuna rao chavala <arjunaraoc googlemail com>:
> Hi Friends,
>
> Pango still has few rendering issues with Telugu and Kannada,
> notably with Akhand (k+sha and its matra variants processing)
> https://bugzilla.gnome.org/show_bug.cgi?id=579398
> https://bugzilla.gnome.org/show_bug.cgi?id=604060
> These were detected over one year back. Even the fix was developed,
> but kept pending as Firefox has rendering problems (for Telugu only)
> with the fix. Now Chrome is available as an alternate browser, I would
> like to move ahead to submit the patch upstream and pursue for
> inclusion in the next release.
>
> I have validated the issue in 1.28. and also verified that the patch
> works for Telugu (sure), Kannada and that there is no regression for
> Hindi (confident). As the Indic module is common to several other
> languages, I need help to test the patch for any regression with
> other languages also confirm my findings with Kannada and Hindi.
>
> I have built Debian packages with the fix for Ubuntu 10.04 LTS
> (libpango1.0-common_1.28.0-0ubuntu3_all.deb,
> libpango1.0-0_1.28.0-0ubuntu3_i386.deb and few other libraries) to
> speed up this process and can share the same. In case of any problem,
> you can easily revert to the previous packages by a reinstall through
> your package manager. Testing is though visiting few websites of your
> language or editing an article using gedit and also checking for any
> akhand combinations of your language. I will provide a test file for
> each of the language for any Akhand issues.
>
> Please reply to this email against the languages that you can test. I
> am looking for at least two people to test for each language.
>
> Devanagari (Hindi) ..Arjun
> Bengali
> Gurumukhi
> Gujarathi
> Oriya
> Tamil
> Telugu ..Arjun
> Kannada..Arjun
> Malayalam
> Sinhala
>
> As Indic language applications on Linux for e-governance and
> e-commerce are increasing, we can't afford to continue to live with
> issues with Indic rendering on Gnome.
>
> Thanks in advance for your cooperation
>
> Best regards
> Arjun
>
> PS: Technical details of the issue: The Akhand needs to be handled
> through GSUB at the character cluster level. It is being done in
> (1.28.x) after the entire glyph sequence is converted for a word/span
> of text. The patch fixes the same.
>