[Smc-discuss] Pango rendering issues

[Vinod]

Hi Raj,
I looked up the new document after your first mail
about this issue. My understanding from reading the
Devanagari and Malayalam sections was pretty much what
you stated below. 

As I see it, it would be desirable if this were not
the case [purely from an implementational point of
view]: Consider the following two typing sequences:

1. PA+AA-sign+LA+virama (for paal with la+chandrakkala
at the end), and
2. LA+virama+PA (for the conjunct lpa)

If LA+virama renders as the chill form [which seems to
be our interpretation], then in the first case we will
need the keyboard to produce the Unicode sequence
LA+virama+ZWNJ by default so that the desired
LA+chandrakkala gets rendered. But this is not a good
idea since we have no idea whether the user will type
a consonant after the virama in which case we don't
want the ZWNJ.

If on the other hand, as Paul prefers, the Unicode
sequence LA+virama is NOT rendered as the chill form,
but as LA+virama, then there is no problem. A user who
wants the chill form will type the key for the chill
[which is available in the current standards for
Malayalam keyboards] which'll produce the required
Unicode sequence with ZWJ.

I don't know whether this concern is really relevant.
If the keyboard + OS interface can be programmed to
change the Unicode sequence produced for the previous
key stroke this wouldn't be a problem. I'm not sure if
this flexibilty is available with current designs.

Yet another concern is that if you allow two types of
Unicode sequences to produce the same rendering, we
will have problems with searching etc. i.e., if
LA+virama and LA+virama+ZWNJ (or ZWJ if you prefer the
chill to be rendered when virama is left hanging)
produce the same glyph sequence, wouldn't we have a
problem with searching?

Thanks,
Vinod


--- Rajkumar S <raj at linuxense.com> wrote:
> Hi,
> 
> I do not know if you are aware of this, but latest
> unicode standards has 
> expanded the chapter on Malayalam has some
> references to chillu. The 
> standards are available at 
>
http://www.unicode.org/versions/Unicode4.0.0/ch09.pdf
> I have updated my 
> page at http://www.linuxense.com/oss/render/ with
> this information.
> 
> I have a small question though, In my discussion
> with Ajaylal yesterday, 
> we had a small confusion. If chillu forms are
> considered equivalent of 
> dead consonant, then consonant + virama will form a
> chillu variant if 
> the consonant can form a chillu, as given by rule R1
> in section 9.1 of 
> Unicode std, 4.0 (p 225) To force the usage of
> chandrakalla variant, we 
> have to use ZWNJ. Is this understanding correct?
> 
> I have included the examples in both utf-8 text and
> pictures.
> 
> paal (1) is the chillu form and palle (2) is the
> chandrakala from of the 
> same word, The figures in bracket point to the
> figures in picture and text.
> 
> Also from the unicode standards,
> 
> "Five sonorant consonants merge with the virama when
> they appear in 
> syllable-final position with no inherent vowel. A
> consonant when so 
> merged is called cillaks.aram:"
> 
> do not seems to be correct as the word, avalkke (3)
> has a chille non 
> syllable-final position. lkke, at the end of the
> word is a single syllable.
> 
> yudit is not rendering the word correctly so I have
> used a space in 
> between to get the word and then deleted the space,
> (a (mis) feature in 
> yudit :)
> 
> The utf-8 text and figures of examples used are
> attached.
> 
> raj
> > 1 പാല്  
> 2 പാല്‌ 
> 3 അവള്ക്ക്‌

> ATTACHMENT part 3 image/png name=text1.png
> _______________________________________________
> Smc-discuss mailing list
> Smc-discuss at lists.sarovar.org
>
http://lists.sarovar.org/cgi-bin/mailman/listinfo/smc-discuss
>