From: Peter Kirk (peter.r.kirk@ntlworld.com)
Date: Tue Jul 29 2003 - 19:37:05 EDT
On 29/07/2003 16:28, John Hudson wrote:
> At 04:11 PM 7/29/2003, Peter Kirk wrote:
>
>> Either I have not made myself clear or my understanding of the
>> rendering process is even less than I thought. Perhaps I should have
>> said "glyph" rather than "character". But the real point is that I am
>> suggesting some kind of flag which could be preserved from outputting
>> on glyph to outputting the next, on the lines of "the last glyph I
>> output was a vowel" or "... a consonant" - with "vowel" or
>> "consonant" defined simply as one of a particular list of glyphs or
>> combinations. Is that possible, or is the rendering engine unable to
>> preserve any kind of state from glyph to glyph?
>
>
> It is possible to store information about a glyph for processing
> purposes. In OpenType this is done in the GDEF table, but the glyph
> types are currently limited to simple, ligature, mark and component.
> It is not essential to make such assignments much of the time; for
> example, you only need to classify a ligature as such if you want to
> position marks relative to different parts of the ligature (in which
> case you also define how many components the ligature has. The only
> GDEF classification currently necessary for Hebrew is 'mark'. If you
> wanted the GDEF table specification extended for, e.g. a distinction
> between consonants and vowels, you would need to approach MS and
> Adobe. I really don't recommend doing that at this stage, since this
> really is a problem that should be solveable at the text encoding
> level. The OpenType philosophy is very much opposed to handling
> anything that looks like a character processing issue in glyph space
> (unlike AAT and Graphite).
>
> John Hudson
>
>
Well, I still think we are talking at cross purposes. I am not talking
about storing this flag in the font but as a variable in the rendering
process so that it how it renders the current glyph depends on a flag
set when it rendered the previous one. To put it another way, the
renderer is not a stateless machine.
But I agree with you that this is a problem that should be solveable at
the text encoding level. I would suggest this kind of solution only if
the encoding level solution is rejected in response to arguments like
"some people nowadays don't make this distinction and we don't want to
confuse them, so this distinction which many people have made for 1000
years should not be encodable in Unicode".
-- Peter Kirk peter.r.kirk@ntlworld.com http://web.onetel.net.uk/~peterkirk/
This archive was generated by hypermail 2.1.5 : Tue Jul 29 2003 - 20:06:46 EDT