Glyph naming uncompliant with AGLFN version 1.5+



  • Disclaimer: I'm not a RoboFont user, but when I have to examine a font, I typically recognize a font created using RoboFont by some bugs in its glyph naming, as if the changes of version 1.5 and later in AGLFN https://github.com/adobe-type-tools/agl-aglfn/blob/master/aglfn.txt hadn't been followed.

    The glyphs in question are (for a static font):

    • dotbelow: according to aglfn.txt, the correct name is dotbelowcomb.
    • commaaccent: accoring to aglfn.txt version 1.7, the correct name is uniF6C3.
    • mu (U+00B5), Omega (U+2126), Delta (U+2206): the standard is a bit misleading here, but the logical consequence of the changes in version 1.5 (and consistent with https://github.com/adobe-type-tools/agl-aglfn/blob/master/glyphlist.txt) is that the names must be, respectively, micro, Ohm and increment (or uni*, see below).

    Btw, I was recently able to compare a font produced by RofoFont + ADFKO, in static and variable version, and, although the static version has all the aforementioned typical bugs, the variable version did have uniF6C3, while the static version had commaaccent. For micro, Ohm and increment it had uni00B5, uni2126 and uni2206 (which is also correct). dotbelowcomb was however still incorrect, since noted dotbelow.



  • According to @frederik said in Are the auto names correct for /guillemotleft and /guillemotright?:

    a guillemot (with o) is a bird... that has been a typo in AGLFN, since forever.

    RoboFont uses GNFUL to map unicodes to glyph names. However you can always set your own GNFUL list, even in the format of a AGLFN, see the preferences

    But U+0323 in GNFUL is dotbelowcmb, still not dotbelow. U+00B5 is mu.math (well, why not…), U+2126 is ohm and U+2206 is increment.


  • admin

    Naming glyph is most of all a personal preference or a preference made by a foundry. AGLFN is one way of naming glyphs, with historical errors and a heavy short sided latinised view (I also understand why AGLFN cannot be changed directly).

    With GNFUL, Erik and I made a package where we tried to make readable glyph names based on the unicode descriptions (uni<hex> is not readable), like:

    • U+0323 COMBINING DOT BELOW --> glyph name dotbelowcmb
    • U+00B5 is not the same as the greek U+03BC with GNFUL we opt strongly to have greek first instead of the math symbol, this is the same for omh ≠ Omega (the greek capital)

    I would not consider these as bugs but as improvement for all greek designers.

    As stated above: setting up glyph names is a choice of a designer: if one finds it suitable to draw uni<hex> that is possible, the same with more descriptive names.

    hope this helps!



  • OK, you favor GNFUL (this is already an improvement, as for example dotbelowcmb exists in https://github.com/adobe-type-tools/agl-aglfn/blob/master/glyphlist.txt and thus will be correctly parsed by Adobe apps like Adobe Acrobat).

    For the Greek, I agree with you: U+00B5, U+2126, U+2206 aren't Greek letters, they're symbols derived from Greek letters and therefore shouldn't receive the same names (and therefore not mu, Omega and Delta).

    But since we're more or less in agreement, why every time I see this very questionable choice of glyph names (dotbelow, commaaccent, mu, Omega, Delta, and hookabove too), the foundry always responds to me: "we created our fonts with RoboFont"? They're using an old version, with an old version of GNFUL, I have to tell them to update?


Log in to reply