Mahesh T. Pai wrote
> PUA isn't necessary, and a font technology that handles elements of
> complex script shaping by referencing PUAs isn't fundamentally any
> different from one that uses glyph names or another identifier and
> leaves the glyph unencoded.
Does one exist? Does it work? (we will leave out the acceptance /
popularity part).
I recall old Arabic layout models that relied on AFII glyph naming, but
there are good reasons why such systems were never widely popular and
have not persisted. There are also custom Indic layout models still in
use that rely on fonts made in particular ways with particular glyph
sets, which I would consider similar in that the layout intelligence
lies outside the font in the software (in this case in plug-ins for
InDesign). Such models tend to be intinsically limited in their
capabilities, because the fonts must contain very particular collections
of glyphs named (or encoded) in particular ways. Smart font formats such
as OpenType, AAT and Graphite, are much more flexible, because one can
solve layout problems in more than one way. For example, I have recently
been working on a Odia (Oriya) font in which I needed vertically
shortened forms of letters in conjunct-initial positions due to
technical limitations of the target environment in which the font is
used. These are accessed using OpenType contextual GSUB lookups. This is
an example of a layout solution that is design-specific, and which would
not be possible if I were working in a layout model with a fixed set of
recognised identifiers.
I don't know why OT Layout is not yet implemented in Android phones. I
can think of a number of possible reasons, a combination of which might
apply. One is that the developers simply have not done the work yet, but
intend to. Another is that they have concerns about font size on mobile
devices, which has delayed support for fonts with large layout tables.
Another is that they have security concerns about OTL tables in fonts
(Google's webfont sanitiser was stripping OTL tables from fonts served
to Chrome for this reason, as I understand; I'm not sure if this has
changed yet).
JH
