[HarfBuzz] [p-c] Perso-Arabic symbols for "year"

[Behdad Esfahbod]

BTW, I came across this picture from six years ago, before Unicode had fixed
the bidi category of the Arabic number sign:

  https://secure.flickr.com/photos/behdad/107995524

I still have no idea how to fix it in HarfBuzz without breaking other stuff...



On 06/04/2012 02:16 AM, Connie Bobroff wrote:
> Khaled,
> Thanks and congratulations. The new Amiri works great on Firefox. All 4 digits
> are nicely above the Arabic Number Sign. (Just note that if the right-most
> digit is "8", the bottom, right portion collides with the Arabic Number Sign a
> bit. But no big deal.
> I think the other points, many of which are critical issues, need to be taken
> up in a separate thread. Please hang on.
> 
> 
>  
> On Mon, Jun 4, 2012 at 11:02 AM, Khaled Hosny <khaledhosny at eglug.org
> <mailto:khaledhosny at eglug.org>> wrote:
> 
>     On Mon, Jun 04, 2012 at 08:26:47AM +0530, Connie Bobroff wrote:
>     > Khaled,
>     > Yes, please send the version of Amiri that works on Firefox. The latest
>     version
>     > from SourceForge works very nicely on IE but not Firefox.
> 
>     File attached.
> 
>     > Amiri is nice in that it's relatively the same size as Tahoma and
>     > therefore, much more suitable for web use than the others.
> 
>     That is by design, almost all Arabic fonts are ridiculously too small
>     compared to Latin for reason beyond me, making it impossible to use the
>     same point size for Arabic and non-Arabic text.
> 
>     > (However, for this particular project, unfortunately, I won't be able
>     > to use it since the Heh Goal is not the right shape to match the old
>     > style manuscript.)
> 
>     This was intentional, IMO many of the variant Arabic letters in Unicode
>     are really stylistic variants specific to certain calligraphic style
>     commonly used for certain languages, Heh Goal is such an example, it is
>     a Nastliq-style variant of the regular Heh and a calligrapher writing
>     Heh this way will do so even when writing Arabic or any other language
>     using Arabic script, the same applies for the so-called Keheh or Farsi
>     Kaf which is really a Nastaliq Kaf, and Arabic calligraphers writing in
>     Nastaliq use that very same Kaf. So for Amiri, being a Naskh style font,
>     I decided that this variance makes no sense and does not match the
>     design so I ignored it, however if there is evidence of manuscripts in
>     Naskh style making that distinction I'll happily follow it.
> 
>     > I did not quite see until now these "Advanced Features" on SIL which say how
>     > many digits the font allows:
>     > http://scripts.sil.org/cms/scripts/page.php?item_id=ArabicFonts
>     > Probably there should be a discussion somewhere about using "Arabic Number
>     > Sign" for dates and therefore needing 4 digits. I presume "Arabic Sign
>     Samvat"
>     > will be given 4 digits but as I said, it's the wrong shape at least for
>     me in
>     > this project.
> 
>     The problem is that I don’t really know what is the intended use of the
>     Arabic number sign, the available Unicode proposals that supposedly lead
>     to its
>     inclusion do not say much about it:
>     http://std.dkuug.dk/jtc1/sc2/wg2/docs/n2413.htm
> 
>     This page says it is “used to indicate the beginning of a number” which
>     does not say much either, but suggests it takes an arbitrary number
>     digits, so I extended Amiri’s number sign to accept 4 digits:
>     http://people.w3.org/rishida/scripts/urdu/
> 
>     Regards,
>      Khaled
> 
>     > -Connie
>     >
>     > On Sun, Jun 3, 2012 at 10:42 PM, Behdad Esfahbod <behdad at behdad.org
>     <mailto:behdad at behdad.org>> wrote:
>     >
>     >     On 06/03/2012 12:19 PM, Khaled Hosny wrote:
>     >     > This have been fixed in FontForge’s master branch, it will break
>     feature
>     >     > files expecting the older behaviour, but there is not much we can do
>     >     > here, and it is hardly the only incompatible fix to FontForge’s
>     feature
>     >     > file code since the last release.
>     >
>     >     Thanks Khaled.  I still don't have any idea how to correctly handle
>     these
>     >     sequences in HarfBuzz, but will experiment with Uniscribe, try to
>     see if I
>     >     can
>     >     figure out what it's doing.
>     >
>     >     behdad
>     >
>     >
> 
>