[cairo] Looks Like Multiple Masters Is A Dead Technology

[Behdad Esfahbod]

On 15-12-11 11:04 PM, Lawrence D'Oliveiro wrote:
> On Fri, 11 Dec 2015 12:04:34 +0100, Behdad Esfahbod wrote:
> 
>> On 15-12-10 08:16 PM, Lawrence D'Oliveiro wrote:
> <
>>> While font-design tools are acquiring all
>>> kinds of parametric, even scripting, capability, it seems like
>>> parametric type rendering has not been very popular 
>>> <https://www.typotheque.com/articles/typeface_as_programme_peter_bilak>.
>>
>> Yes, parametric type design is not popular.  That's why we are
>> building GX from workflows that font designers are already using:
>> multiple compatible masters.  That's different from parametric design.
> 
> That’s the other way round from the article I quoted.
> 
> I remember TrueType GX was criticized for including full finite-state
> machines in the font format for implementing variations and
> context-dependent kerning and substitutions.
> 
> Whereas I gather the philosophy of OpenType was to build the smarts into
> the rendering library, not the font format.

There are two different things:

  - Glyph variations: Multiple Masters, and TrueType GX's fvar/gvar/avar:
these allow different instances of the font (different weight, stretch,
optical size, whatever) be instantiated from multiple masters at runtime.
OpenType doesn't address this.  I, and a few others, are working on bringing
this aspect of TrueType GX to OpenType.

  - Glyph substitution and positioning: OpenType has GSUB / GPOS while AAT
(what TrueType GX got renamed to) has the state-machines, ie. morx / kerx.
While I'm also interested in bringing these to OpenType, that's different from
what this thread was about.


> So do you feel you want to move away from that philosophy in OpenType
> 2.0?

Not move away per se, but give font designers more options.

behdad