i = dotlessi + dotaccent

And j = uni0237 + dotaccent.

What could go wrong?

Comments

  • John HudsonJohn Hudson Posts: 1,415
    In what sense?

    Are you proposing to build i and j as composites, as described? That's fine.

    Are you proposing the encode i and j as described in text? That's not fine, because i and j do not have canonical decompositions.
  • My guess is that Nick just switched to Glyphs and discovered that i and j are built like this by default.
  • Florian HardwigFlorian Hardwig Posts: 139
    edited October 2015
    … which helps to avoid a common error made in regard to the dot accent:

    “The dot symbol (=overdot) should be placed and shaped the same as the dot above i” — diacritics.typo.cz
    “[The Polish kropka in ż] should be always identically shaped and placed as the dot above i.” — Adam Twardoch
    “We have words like ‘išėjimas’ and it’s crucial that all the dots are the same.” — Lithuanian typographers at 6 pt conf.

    It’s pretty simple, but there are still too many font makers who are not aware of this.
  • Cool, something I have been doing right. :)
  • It’s pretty simple

    It's no so simple in heavy fonts where a dot the same size as the tittle can look huge relative to other diacritical marks. If other accents have to work with a massive dot accent they can end up too wide.

  • Nick ShinnNick Shinn Posts: 1,284
    Yes, as composites.
    Still working in FontLab, just pursuing efficiency.

    I concur with James’ observation about the heavy weights—might we just follow the example of Gill Sans Extra Bold for a tiny tittle? 
  • It's no so simple in heavy fonts where a dot the same size as the tittle can look huge relative to other diacritical marks.

    Right, heavy styles are tricky. Still, it’s only true as long as you regard the tittle as something different from the dot accent. The i/j dot needs to harmonize with (other) diacritical marks as well, no?
  • Nick ShinnNick Shinn Posts: 1,284
    A reminder: don’t flip or rotate your components.
  • Unless you flip them twice. That's ok.
  • The i/j dot needs to harmonize with (other) diacritical marks as well, no?

    No, the tittle needs to harmonize with the stem of i, and the dot accent needs to harmonize with other accents. For example:


  • Or only flip them vertically, right?
  • Does it need to match the width of the stem or do we think it needs to? I've made fat fonts where the x height is so damn high that the lowercase i ends up looking like a lowercase L. To maximize the stem/tittle gap, I can drop the stem down a bit or pop the tittle over the ascender height. But if I'd used a dot accent I wouldn't have had to resort to such sorcery. Did Nick just trigger the end of the tittle as we know it? I hope so, because I'm tired of saying tittle.
  • Still, it’s only true as long as you regard the tittle as something different from the dot accent. The i/j dot needs to harmonize with (other) diacritical marks as well, no?

    But "harmonize with" doesn't mean "be identical to".  A dot accent that must look good over the rounded top of an e can seem a bit starved when it appears over the stem of the i.  And in black weights, a dieresis made of two tittles can look huge.  If you use the same component for all these, you don't get to adjust them optically.
  • Kent LewKent Lew Posts: 763
    Flipping twice = rotating. Assuming “flip twice” means flip horiz then vert (or vert then horiz). Either way, path direction stays constant. Not so with a single flip.

    [BTW, back in the day, when I was coming up as a paste-up artist, we used to distinguish between “flip” (meaning rotate) and “flop” (meaning what we think of now as flip).] 

  • Nick ShinnNick Shinn Posts: 1,284
    edited October 2015
    Polish test word for dotaccent theory: dążenie

    ***

    Mark—love your Seussy poesy!


  • Heh, Fox In Socks was one of my favorite books to read to my daughter when she was little. I had to do it.
  • No, the tittle needs to harmonize with the stem of i, and the dot accent needs to harmonize with other accents.

    The e with dot (ė) is pretty much exclusive to the Lithuanian spelling. From what I’ve heard and seen from native typographers, they’d consider the example shown above as erroneous, and would prefer something like this:

    Lithuanian uses neither a diaeresis, so no direct need to harmonize with that, nor an e with macron. The dot may be round, square or diamond-shaped, square or oblong – you name it – it’s all fine as long as it is (visually) equal to the dot on ‘i’.

    Judging from the work of Łukasz Dziedzic and other native type designers, the preferences of the Polish in regard to the z with dot (ż) are the same. Here is FF Clan Ultra:


    I haven’t heard reliable statements from Maltese typographers, but I’m curious to hear from them. The Irish use of the dot accent in ḃ ċ ḋ ḟ ġ ṁ ṗ ṡ ṫ will require a different solution for heavier weights than the Lithuanian ė and the Polish ż. So does the dot in the Catalan ela geminada (ŀ).
  • edited October 2015
    @Florian Hardwig There is also ė in Samogitian (the Samogitan area is part of Lithuania, though), Kazakh Latin, Ulithian, Tajik (Cyrillic) Romanization and probably more that I don’t know of. There’s more tittles to be found in Asturian, Extremaduran, Livonian, Sanskrit transliteration, Gujarati romanization, Bolognese (though, to be strict, not much in the way of good sources), Javanese, Marshallese (MOD orthography), Nepali (Devanagari) romanization, Ukrainian Abecadło, Old English, Iñupiaq and most of the smaller Polish languages.
  • edited October 2015
    In Asturian, Extremaduran, Bolognese and Livonian, the i-less-tittle (whom I presume the arch nemesis of the dotless i) also coexists with the dieresis.
  • Flipping twice = rotating. Assuming “flip twice” means flip horiz then vert (or vert then horiz). Either way, path direction stays constant. Not so with a single flip.

    In Glyphs, the path direction is corrected if the component is flipped. 

  • In Glyphs, the path direction is corrected if the component is flipped.
    Georg always has my back.

  • Kent LewKent Lew Posts: 763
    In Glyphs, the path direction is corrected if the component is flipped.
    You mean when/if it is decomposed. As a referenced component, the outline direction will necessarily be reversed, won’t it?

  • It will not touch the outlines of the base glyph. On only the paths of the component (e.g. When decomposing or simply drawing them in the app) are reversed.
Sign In or Register to comment.