Advanced Glyph Reordering

I just wrote a blog post about a technique for reordering glyphs using OpenType Layout which I think deserves to be better known. Comments are welcome, especially if they're kind!

Comments

  • k.l.
    k.l. Posts: 109
    Which leads to two (rhetorical) questions:
    Should reordering be up to the font in the first place?
    Why use OpenType Layout for something it evidently hasn’t been made for (I could not have come up with a nicer illustration than the one you provided here), only to boast being able to work around that with some super duper glyph reordering gymnastics?

    Turning to something more trivial:
    That same introductory rant, albeit without some of the more fancy complications, applies to dumb Latin script fonts too. Why add all the precomposed diacritic letters in the first place (unless where design asks for it)? Because even rather simple functionality that OTL offers, like one-to-many substitution, cannot be used, in fonts that are meant to be cross-platform and cross-application compatible, as Microsoft Office does not apply such advanced stuff by default with a script as simple as Latin for which precomposed diacritic letters should good enough. Relying on decomposed glyphs solely, as per your rant (with which I have agreed more a decade before you uttered it), would allow doing away with quite some clutter in fonts ...
  • Simon Cozens
    Simon Cozens Posts: 740
    k.l. said:
    Should reordering be up to the font in the first place?
    No, absolutely not; it should be done by the Universal Shaping Engine. But the Universal Shaping Engine doesn't do what we need here, and that part of the toolchain is out of our control, so we need to do something to work around it.
    k.l. said:

    Why use OpenType Layout for something it evidently hasn’t been made for (I could not have come up with a nicer illustration than the one you provided here), only to boast being able to work around that with some super duper glyph reordering gymnastics?
    Because it's what we've got, and we want to make fonts that work.
  • John Butler
    John Butler Posts: 290
    edited July 2023
    This thread prompted me to read the latest AFDKO documentation, and it gives me the impression that one-to-one and many-to-one substitutions are possible, as they have been from the beginning, and one-to-many substitutions are still not possible over two decades later. Is this correct? Or is it at least correct for AFDKO-generated fonts? (Also, in this life I only care about Latin script.)

    I feel dumb for asking this question but increasingly dumber for not asking.

  • Simon Cozens
    Simon Cozens Posts: 740
    No, "multiple substitutions" are possible and have been for a very long time. See 5.b.
  • John Butler
    John Butler Posts: 290
    OK, thank you, clearly I hallucinated the contents of the same exact website when I read it.
  • Thomas Phinney
    Thomas Phinney Posts: 2,883
    What is not *directly* possible is many-to-many substitutions.
    The same effects are generally achievable through multiple substitutions, mind you.