Re: I have a question :)

> Begin forwarded message:
>
> Date: Wed, 22 Mar 2006 17:05:48 +0800
> From: LingNing Zhang
> Subject: I have a question :)

I have eleven answers :-)


> I want to fix some bugs about Malayalam. But I don't know the
> Malayalam. And I don't know whether the malayalam conjugates  in
> below link  are right or wrong.
> http://bugzilla.gnome.org/attachment.cgi?id=19849&action=view
> Who can help me?

Time.
Malayalam rules are not completely cast in stone, particularly if you try to
involve things like cillu or ZWJ. It is very unfortunate, but it goes like
that...


In details:
1 is _probably_ correct at the end of a word (still debated); if internal to
a word it is even more debated... Alternate is to use plain candrakkala, not
cillu. Be prepared to adjust your code.
2 is officially correct; 3 left also official; 4 too.
3 right is very controversed: I am of the idea it goes this way, but I
believe I am in minority about this one. Watch out.
5 is wrong.

8 is triply wrong. lla is a well-known conjunct; secondary la has a
well-known subscripted form; if you want to use cillu l, it has to comes
*before* (what you have written is read "lal").
9, 10 and 11 suffers form the same symptoms: there is a well-known conjunct
for them, there are secondary forms (for ?va, you got the one for ?ya), and
the order is reversed.

14 and 15 is under discussion. Watch out. My personnal guess is that 14
shall be correct, and that 15 should render the same as 14.

18 and 19 are wrong: secondary forms takes precedences over use of cillus in
conjunct. Should be plain Na or plain na followed by the other form of ya
(the one you got on line 10).

22 is wrong (what you wrote is read "nar"); it should look like the
secondary form of ra (not a cillu) *before* na; this reordering is an
exception to the rule used for the other secondary form (?ya, ?la, ?va).
23 is not firmly defined (but what you show is wrong: at the very least it
misses a candrakkala). It is still unclear how to encode the well-known
conjunct (nTa) you wrote; my idea is (the natural combination)
0D28+0D4D+0D31; others want to insert ZWJ in the sequence (or special
codepoints for the cillus); there is a related problem about how to encode
the unstacked version.

26 is incorrect, as you pointed out: e sign should go in the middle of the
two other glyphs.


Hope this helps,

Antoine