Fwd: Malayalam in Pango

---------- Forwarded message ----------
From: Arun M <arun-0jIIvIziipmf0DUV/[email protected]>
Date: Apr 3, 2006 3:58 PM
Subject: Re: [gnu-malayalam-discuss] (no subject)
To: [email protected], Vimal Joseph <[email protected]>


Hi,

  I just found your mail. Here are my comments regarding rendering
issues. There are some Malayalam related issues not covered in your
attachment. I will get back to you on that later.

While working on Malayalam please not that most of Malayalam unicode fonts =
have
wrong OTF table. Rachana looks to be the best so far.

Mahesh had replied correctly on most of rendering issues.


> I have no comment on line 5; this is a result of the ambiguity in the
> unicode standard as shown in line 1 and 2.

I think line 5 rendering should be <na><virama><ma>

> Line 10 - it is the ``virama + character'' which should be substituted
> and placed after the first character. This line shows the substituted
> glyph being placed before the first character, or the ``character +
> virama'' sequence being substituted, and therefore wrong.

Rendering should be either <ya><virama><ya> or ligature for <YA>.
ya has post base form.
See the section of PSTS
http://www.microsoft.com/typography/otfntdev/indicot/features.aspx
Example shown is that of another consonant <va>. <ya> and <va> acts similar=
ly.

But ICU/Pango inherits a mistake in Microsoft documentation here
http://www.microsoft.com/typography/otfntdev/indicot/appen.aspx
hence rendering is different.This document says <ya> has Post base
form and va  none. Which is wrong, Ya and Va behave  similarly.


> Lines 18 and 19
>
> ``C1+ virama + C2'' should substitute the conjunct of C2; not C1.

Line 18 and 19 should be <C1> <Post base form of Ya> Or ligature for
< C1+Post base form of Ya>


> Line 22 is wrong; The correct glyph substitution in a C1 + virama + C2
> sequence is to show the conjunct glyph if available, or substitute the
> conjunct glyph or the virama + C2 sequence, or show the explicit virama.
>
> Line 23.  hahahahahahah!!!! There should be a zwj in the sequence,
> immediately after the virama. IMHO.

22 and 23 I will comment later,

> Line 26 does not parse for me. IMHO, the rendering for the sequence
> shown is correct.

This is wrong,

In the case of <c1> <virama> <c2> <vowel sign>

If the font has the ligature for <c1> + <virama> + <c2> then
rendering will be  <vowel sign> <c1+virama+c2>
else rendering will be <c1><virama><vowelsign><c2>


Please mail me if you need more details.

If i remember correctly one bug I find in pango is, after reordering
when glyph substitution starts
even characters from two different clusters are allowed to combine
together to form a glyph.

with regards,
arun.


--
Free Software, Free Society <http://fsfs.hipatia.net>