[omega] Question about the paper published in EuroTeX 2005

[Chris Rowley]

Yannis

> > More to the general idea that, as I understand it, one is not meant to
> > access glyphs in an OpenType-style of font resource
> 

Sorry, that sentence did not get finished: it should go on:

...via the glyph index directly, even when one knows exactly which
glyph you want and where to put it.

> That's a very interesting issue we are dealing with. technically 
> speaking, the glyph index is not a robust way to access a glyph in a 
> TrueType-like structure. For example, you can re-order the glyphs in a 
> TrueType font and it still is exactly the same font (while Type 1 fonts 
> have glyph names).
> 

Exactly: and that is the problem, no glyph `names'!

My main point is that this should be seen as an important
implementation issue when using this particular font technology, not
as a reason for building `glyph access via Unicode slots numbers' into
the model.

> But we don't have the choice. If we want to interfere with OpenType 
> tables we just have to dig into the font and access the glyphs. Maybe 
> we will use signature calculations to find unique identifiers for 
> individual glyphs in a font so that we can recognize them if for any 
> reason the font is re-ordered. That would require a table saying that 
> "glyph index 37 = glyph signature 1F35OA726BB89A" and then one would be 
> able write code using \glyphindex{37}.
> 

Yes, something like that is a possible interface.

> This brings us back to the old problem about how do we identify a font? 
> (besides version of the whole, of each glyph, of metrics, of metadata, 
> of encodings, one could use signatures, and even the DNA signature of 
> the designer).

Now there I really did not want to go...!!

But at least we now have, at the implementation-level even, the idea
of a font resource which could contain the designer's DNA (and a
measure of his mood as he designed each letter).  But it is not, as
yet, very good at making its resources avialble to the typesetting
software.

> 
> To return to OpenType fonts, the current status of our development plan 
> is to implement "consulting of metrics", "consulting of GPOS" and 
> "consulting of GSUB" as three actions similar to OTPs, so that one can 
> place real OTPs between or after these actions.

Good; I get the idea.  What metrics does a typical OT font contain?


chris