[HarfBuzz] Tai Tham Shaping Question #2 : MEDIAL RA

[Behdad Esfahbod]

On 12-11-14 08:36 PM, Theppitak Karoonboonyanan wrote:
> On Thu, Nov 15, 2012 at 2:53 AM, Behdad Esfahbod <behdad at behdad.org> wrote:
>> On 12-11-14 01:15 AM, Theppitak Karoonboonyanan wrote:
>>> On Tue, Nov 13, 2012 at 4:45 AM, Behdad Esfahbod <behdad at behdad.org> wrote:
>>>> Ed!
>>>>
>>>> There's no way we can fix these without a font!  Just send the font first
>>>> already :).  Jonathan and I are having yet another HarfBuzz Hackfest week this
>>>> week and like to address the remaining issues for "Tai *" scripts :).
>>>
>>> While waiting for Ed, you may try Khottabun font for Lao/Esaan Tham script,
>>> another Tai Tham variation used in Lao and northeastern Thailand:
>>>
>>>  http://linux.thai.net/~thep/esaan-scripts/fonts/Khottabun.ttf
>>>
>>> The font, however, has provided all the reordering rules in GSUB, so that it
>>> can still display on absence of rendering engine preprocessing. It may not be
>>> good for your testing, but you can get some idea of the requirements from it.
>>
>> Humm.  Then it's not interesting indeed.  Even worse, those rules may conflict
>> with the engine processing :|.
> 
> I know. That's why I said it might not be good for your testing. Just hope to
> provide some sample for comparison using another system without the
> processing.

Our current recommendation is that if your font does NOT rely on the engine
reordering, it's best to use the DFLT script.  In HarfBuzz we use that clue to
decide whether to use the Indic shaper or the generic shaper.  In some cases,
using a font designed for the generic shaper with the Indic shaper can produce
inferior results.


> And I hope the sample text in the test page could still be some help.
> 
> One common case that's still missing in the test page, though, is "ᨯᩪᩕᩣ",
> where MEDIAL RA has to be shifted left by two positions.
> The sequence is:
> 
>   U+1A2F TAI THAM LETTER DA
>   U+1A6A TAI THAM VOWEL SIGN UU
>   U+1A55 TAI THAM CONSONANT SIGN MEDIAL RA
>   U+1A63 TAI THAM VOWEL SIGN AA
> 
> And the rendered sequence should be:
> 
>   U+1A55 TAI THAM CONSONANT SIGN MEDIAL RA
>   U+1A2F TAI THAM LETTER DA
>   U+1A6A TAI THAM VOWEL SIGN UU
>   U+1A63 TAI THAM VOWEL SIGN AA

Right.  This behavior is like the medial Ra in Myanmar, so should be fine when
we implement that.


> Regarding the font development, since we want to make progress on other
> parts of the language support, such as input method, adding GSUB is all
> we can do while the rendering engine supports are still lacking.
> 
> Once HarfBuzz supports it, I'll remove the rules and distribute two
> different versions of the font, one for HarfBuzz, the other for other
> engines.

Or just ship one that has both a DFLT table, and a non-DFLT table.


> Regards,
> 

-- 
behdad
http://behdad.org/