bug#11075: 24.0.94; Arabic character composition

[Stefan Kangas <stefan@marxist.se>]

Joost Kremers <joostkremers@fastmail.fm> writes:

> Although Emacs 24 has bidi support (yay!), character composition for Arabic
> does not work for me on OS X. (In the Arabic script, each character has
> different forms depending on its position in the word: beginning, middle or
> end. These shapes should be selected automatically by the software, but
> they are not.)

It looks to me like this works on macOS now.  At least the Arabic
characters compose differently depending on where they are in the
word. Admittedly, my knowledge here is superficial at best, having done
only a brief foray into learning beginner Persian.

So, I can't say for sure if they compose correctly, just that they
compose differently.

Alan Third <alan@idiocy.org> writes:

> On Mon, Sep 25, 2017 at 07:17:21AM -0700, John Wiegley wrote:
>> >>>>> "AT" == Alan Third <alan@idiocy.org> writes:
>>
>> AT> Hi, does anyone know if this bug was resolved? I assume if it worked in
>> AT> the Mac port it will now work in the NS port?
>>
>> Why can that assumption be made now?
>
> In the seven or eight years since this bug report was made, we
> switched over to the mac port font backend.

Since that was written, harfbuzz was enabled by default.  And wouldn't
that library have good support for this anyhow?

> It might still be an issue on GNUstep, though. I think it still uses
> nsfont.m.

I'm not a GNUstep user so I can't easily test this.

But would the recent harfbuzz support (enabled by default on GNUstep
too, I assume) fix this?  Do you know, Eli?

Best regards,
Stefan Kangas