Re: Composing Hebrew diacriticals

[Kenichi Handa <handa@m17n.org>]

In article <AANLkTin5fDs_thjY2I3S4anUDpgkqbzfZt2TBdzRo-8Y@mail.gmail.com>, Yair F <yair.f.lists@gmail.com> writes:

>>> 1. Letter Shin (U+5E9) is not composed. It is followed by Shin-dot
>>> (U+5C2) which breaks the composition.
> >
> > In my environment, the sequence U+5E9 U+5C2 is surely
> > composed.  Please show me the output of this command.
> >
> > % MDEBUG_FLT=3 emacs --eval '(message "\u05E9\u05C2")'
[...]
> Mine is
>  [FLT] (hebr-ff (dejavu sans)
>  [FLT]   (SOURCE 05E9)
>  [FLT]   (STAGE 0 "H" (05E9)
>  [FLT]     (SUBPART 0
>  [FLT]      (COND
>  [FLT]       (REGEX "^." "H" 1
>  [FLT]        (COPY 0x5E9)))))
>  [FLT]   (RESULT (053D 1152 0 0)))

Are you using the same setting as mine which I wrote as
below in the previous mail?

> I have not yet committed proper codes for Hebrew
> composition.  I'm now testing with this simple version.
> 
> (let ((pattern "[\u05D0-\u05F2][\u0591-\u05BF\u05C1-\u05C5\u05C7]+"))
>   (set-char-table-range
>    composition-function-table '(#x591 . #x5C7)
>    (list (vector pattern 1 'font-shape-gstring)
> 	 ["[\u0591-\u05C7]" 0 font-shape-gstring]))
>   (set-char-table-range
>    composition-function-table #x5C0 nil)
>   (set-char-table-range
>    composition-function-table #x5C6 nil))

>>> 2. Letter Lamed (U+5DC) Followed by Holam (U+5B9) The point is
>>> composed to the right of the letter instead of left.
> >
> > It seems that "dejavu sans" font doesn't have 'ccmp' nor
> > 'mark' feature applicable to that sequence.

> In that case, may I suggest switching to a different font?

It's possible to write such a code (which will slow down the
display), but if you don't have a better font, we anyway
have to use the available font.

By the way, do you have a better font than 'dejavu sans' for
Hebrew?

>>> 3. Letter Yod (U+5D9) composed with Hiriq (U+5B4) is composed into
>>> presentation form (U+FB1D). This should only happen with specific
>>> control (Either CGJ or ZWJ I'll check).
> >
> > I tried to show that sequence with gedit and and firefox.
> > Both show it with U+FB1D.

> Kate show this as well, Firefox doesn't. Nevertheless it's still a bug.

Then what is the correct rendering of the sequence "\u05D9\u05B4"?
Applying only `mark' feature?  With dejavu sans font, U+FB1D
has lower dot above baseline, but just applying `mark'
feature results in displaying dot below baseline.

See the attached image.  The left one is the composition
with `ccmp' GSUB feature (i.e. into U+FB1D) for the sequence
"\u05D9\u200D\u05B4", and the right one is the composition
with `mark' GPOS feature for the sequence "\u05D9\u05B4".
Is this the correct rendering?

---
Kenichi Handa
handa@m17n.org