[Debian-in-workers] Fwd: [FTC] Re: Bugs in ttf-tamil-fonts

Santhosh Thottingal santhosh.thottingal at gmail.com
Mon Dec 12 08:46:25 UTC 2011


---------- Forwarded message ----------
From: கா. சேது | කා. සේතු | K. Sethu <skhome at gmail.com>
Date: 2011/12/12
Subject: Re: [FTC] Re: [Debian-in-workers] Bugs in ttf-tamil-fonts
To: freetamilcomputing at googlegroups.com
Cc: Vasudev Kamath <kamathvasudev at gmail.com>, Santhosh Thottingal
<santhosh.thottingal at gmail.com>


On Fri, Dec 9, 2011 at 4:13 PM, Santhosh Thottingal
<santhosh.thottingal at gmail.com> wrote:
> Adding Free Tamil Computing group in CC for their comments.
>

My opinions here in this response need not be the opinion of our FTC group.

Also I did not include Cc to "Debian-in-workers" mailing list since I
am not member in it. But I request you (Santhosh) to please forward
this to them too because I have filed in today a Debian bug report on
a related matter which I mention later below.


> IMO, there no reason to keep TsCU and Tamu fonts in package, since
> they use ASCII font encoding(using ascii code points and tamil glyphs
> for that).

Not only in the ASCII (Latin) range but characters / ligatures are
strewn around in some other key outer ranges also (in addition to the
Tamil Unicode range).

Actually more problems arise from the use of Unicode range punctuation
symbols such as ellipses (U+2026) and other punctuation symbols near
that in TSCu_Paranar. The upstream Fontconfig xml tree comes with  a
configuration file 65-nonlatin.conf that includes a matching
preference of TSCu_Paranar for Sans at the higher priority than "Lohit
Tamil" .

Due to this, while operating in Tamil locale, problems have been seen
in applications like Firefox / Gmail interface wherein symbol used for
left or right arrows are rendered with characters picked from
TSCu_Paranar. Also ellipses symbol used to mean continuation in tabs
of minimised apps in desktop, appearing as ஸ because TSCu fonts use
U+2026 for ஸ.

I can post later screen-shots of such bugs (which were never brought
to the notice of i18n teams of the different Linux distros) and write
an update on the status in current releases.

Actually both TSCu and TAMu fonts use such outside codepoints in
addition to the codepoints for the same characters in the respective
ASCII range (for TSC or TAM encoding) as well as Unicode Tamil range
(for Unicode encoding). I guess the reason is that only for those
characters,  the ASCII code points coincide with Unicode Control
Characters and so additional code points from other Unicode ranges are
used to tackle any rendering problems in using code points of control
characters.

In fact one can see from the common IME keymaps for TSCII, TAM/TAB
(such as in NHM-Writer for Windows)  map such characters which have
ASCII code points of control characters to codepoints in the
Punctuation and other ranges!  All such keyboards and fonts seem to
consistently use same set of codepoints for such characters which
could only mean some standards are in existence for using alternative
codepoints from other ranges for such characters having code points of
control characters in ASCII range?

Whether or not TSCu and TAMu fonts are continued in the package what
should be done is not to provide any priority for hybridized fonts
such as these in the Fontconfig. So the upstream Fontconfig project
also needs to be pointed out to not include TSCu_Paranar (and any
other TSCu or TAMu type hybridized fonts) from the list of preferred
fonts in 65-nonlatin.conf file!.

At this point let me point out another set of problems encountered in
Debian and Debian based distros (including Ubuntu) of current and
recent past. The font matching for Indic scripts have become messy due
to configuration files numberd "90" for Indic scripts in the xml tree,
introduced about a year or so. You can see my bug report and a follow
up mail at : http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=651768

>FTC has recently started some work on Thene, Thendral ,
> Vaigai fonts and we can consider adding them to Tamil Fonts package.
>

Our FTC / Thamizha group effort is temporarily stagnant. I wish to
continue soon but on late Umar's fonts - Thendral, Thenee and  Vaigai,
we (the Thamizha group mebers) will have to discuss if there is need
to make our own modifications because just before we had started in
October, the ICTA (Information and Communication Technology Agency of
Sri Lanka : http://www.icta.lk ) agency  had also released their
derivative version of the same fonts (I learnt about them only a few
weeks back!).

See:
http://www.icta.lk/en/programmes/pli-development/104-local-languages-initiative-/1094-stylized-tamil-unicode-fonts-for-windows-linux.html

The 6 fonts they had released are unicode only and stylized versions
of the  above mentioned 3 Umar fonts and the 3 TSCu fonts (TSC
stripped). Each of them they have named with prefix of "Chemmozhi"
They are released under GPL2 and so merits attention of Web fonts
developers also Linux i18n teams.

Of course more user testing and experience reports with the ICTA's
Chemmozhi fonts also are needed since they are derivative works from
existing fonts and also only recently released. I request Thamizha
group members to test them and write if any problems are encountered.

> The ELCOT font has licensing issues -
> http://elcot.in/tamilfonts_download.php, may be FTC members can
> clarify on that.
>

Yes adding to what others wrote their, just looking at the concerned
font -elcotuni.ttf under Fontforge, I see it does not have any license
statement but its copyright statement is clear :

//Copyright 2007, ELCOT. All rights reserved. //

That itself is too restrictive to be considered for inclusion in
packages (regardless of what their intention was to be).

Further looking at the font under Fontforge I see that it also has
many tamil glyphs in the Latin (ASCII) range. So if we consider  TSCu
/ TAMu fonts not suitable for such reason then elcotuni.ttf also is in
the same boat.

~Sethu



More information about the Debian-in-workers mailing list