>
> I see that you have now changed the hanifi-rohingya as well, but I'm
> not sure that does what you want.  Please consider the effect I
> described above, and decide whether you want 0 or 1 there.


You are right, adding 0 there results in incorrect composition, I have now
reverted my change.
Thanks

On Tue, Sep 27, 2022 at 10:06 PM Eli Zaretskii <eliz@gnu.org> wrote:

> > From: समीर सिंह Sameer Singh <lumarzeli30@gmail.com>
> > Date: Tue, 27 Sep 2022 21:23:40 +0530
> > Cc: 58111@debbugs.gnu.org
> >
> >  This doesn't look right: why 1 and not 0?
> >
> > I copied this from my hanifi-rohingya composition rules, where I do not
> remember
> > why I added 0 instead of 1. Even though they were working as intended I
> have replaced it with 0
> > there just to be safe.
>
> In hanifi-rohingya, the codepoints that triggered composition were
> between #x10D1D and #x10D27, whereas the regexp was
> "[\x10D00-\x10D27]+", which matches many more characters.  With 1 as
> the 2nd element of the rule vector, this means the sequence of
> characters U+10D00 U+10D00 will not cause composition, but the
> sequence U+10D00 U+10D1D will.  By contrast, in Adlam case, the
> characters that trigger composition and the regexp are identical, so I
> see no reason to trigger the composition only on the second character
> of a sequence.
>
> I see that you have now changed the hanifi-rohingya as well, but I'm
> not sure that does what you want.  Please consider the effect I
> described above, and decide whether you want 0 or 1 there.
>