I, for one, prefer having dedicated glyph slots for differently connected forms of a letter. For instance, these are the /fehDotless-ar.medi and /fehDotless-ar.init from my Quinoa Black:
I've had to apply some trickery to the former to make it look like a circle sitting on top of a continuous baseline, when in fact the counter dips below the baseline's top and the circle is optically adjusted to avoid blotting. For the latter, of course, much of that adjustment was unnecessary or would have been counterproductive on its right side.
You can't do this sort of thing if you use the same glyph for connecting and non-connecting instances. Note that Quinoa is a geometric typeface whose Arabic has even been called «more progressive» (as in non-traditional) than its Roman. The vast majority of Arabic typefaces are going to be less geometric, more organic and thus even much less amenable to this kind of treatment than Quinoa.
Regarding nastaliq OpenType Fonts (as distinct from formats that are limited to particular software), I think the Noto Nastaliq Urdu is one of the best, in terms of going to considerable lengths to handle dot interaction within the clumsy mechanism of OpenType contextual GPOS (but also cheating a bit, by using smaller dots than are conventional in Urdu style of nastaliq). But there are still many issues in the interaction of spacing and mark positioning, and the deep problem is that OpenType Layout was not designed to be able to handle cascading styles of Arabic (I have this acknowledgement directly from the man who invented OpenType Layout).
For more information on the Aldhabi project, including illustration of the relatively small glyph set and some of the GPOS issues referred to above, see the slides from my 2012 ISType presentation: Adventures on the way to curls. At the time of the project, we didn't realise just how unsuited to what we were trying to do was the OpenType GPOS mechanism. The result is that the font is really only useful for headlines and other short pieces of display typography, and preferably in environments where the user can manually adjust spacing between lettergroups.
For a more general discussion of the issues around spacing and mark interaction, see my TypeCon 2014 presentation text and illustrations: Problems of adjacency.
Here is an attempt to make a systematisation of the local forms in Bulgarian, Serbian and Macedonian compared to Russian (or better say – traditional Cyrillic script). I use the font Vollkorn for the presentation of the local features. Please do notice that in Bulgaria are still used both the modern form of Bulgarian Cyrillic script (shown in Table 1) and the traditional form (which is same as Russian Cyrillic or traditional Cyrillic script).
Please notice that in my oppinion the italic form of Macedonian Cyrillic „г“ (uni0433.loclMKD) is still in a process of standardisation. It is influenced by the Serbian form on one hand and on the other hand it is influenced by the Bulgarian form. But if we look at the italic form of Macedonian Times – a font prepared by Macedonian designers obviously in the early 90th of the 20th century (the font is not in a Unicode standard) – the „г“ is traditional italic with macron above. The same could be said for the Macedonian Cyrillic „б“. According to Lasko Dzurovski the regular form of Macedonian Cyrillic „б“ could have two forms (traditional and Serbian one – see Table 2), but the italic form of Macedonian Cyrillic „б“ could have only one form (which corresponds to the Serbian one – see Table 3).
@Hrant H. Papazian It will ultimately depend on layout engine support, but ‘optical size’ is a defined standard axis, so I think support will eventually be pretty good.
All the type designer has to do is to map point sizes to the desired axis positions.
I think it will work well for ‘normal’ viewing conditions (computer displays at normal reading distance), but factoring in a variable view distance or even zooming may be tricky to get right for the layout engine programmers.
You're facing one big problem: FF is hardly maintained (and even harder to maintain I hear) and filled to the brink with weirdness and bugs. I have given up on it. Trufont is too early in development though... Regarding interpolation, you can try to export your files in the UFO format and use designspace files and fontmake to interpolate. See e.g. https://github.com/productiontype/Spectral for an example. It's more cumbersome than getting a nice preview in a font editor, but it's better than dealing with FF's quirks.
I don't think it makes sense to have the typical Russian etc. cursive form of г with acute in Macedonian typography if the localised form is used for г without acute. As far as I know, the information provided in the Noto GitHub link that Michel posted above is reliable. The 'locl' feature substitutions for Macedonian should be extended to cover ѓ.
I'll be making this revision to the Brill fonts in the next update.
I don't have information indicating that ѓ is used in any alphabet other than Macedonian; however, I would be wary of making the localised italic form of this glyph the default, just in case it does show up anywhere else. Also, in case of failure of the 'locl' feature, it wouldn't be good to end up with Macedonian ѓ and Russian г — when failing, better to fail consistently.
Can't good hinting make the spacing acceptable based on PPEM?
Not any more. The environments that ignore hints to advance widths are too many and too pervasive. We lost that ability around the time that sub-pixel rendering with sub-pixel positioning became the dominant model.
With regard to the design of Spectral, what I'm mostly struck by is that it seems a nice, traditional book face, in terms of its proportions and style. That strikes me as an odd choice in something that is ostensibly designed for online document use. Some of the illustrations of it in use remind me of pre-Calibri/Cambria MS Office documents, and can't help seeming old-fashioned as a result.
[Disclosure: I suggested to Dave Crossland a few years ago developing a suite of fonts for Google Documents, which would have included both serif and sans, as well as a condensed face especially for spreadsheets. Would still like to do it.]
In the period in question, children in German-speaking countries typically learned two different scripts at school, “Lateinisch” (roundhand) and “Deutsch” (Kurrent, a cursive form of Fraktur). These were used side by side, for different purposes. In a nutshell, Kurrent was the default, while “Lateinisch” was used for foreign languages, loan words, but also proper names and other sorts of emphasis. Unsurprisingly, hybrid forms were quite common.
In this roundhand manuscript, the writer introduces the form of ‘d’ that you describe as oldstyle, and which is associated with Kurrentschrift, as a stylistic alternate. There were no rules for this as with ſ/s. The ‘d’ without downstroke simply lends itself more to being used at the end of words, where it can end in a fancy loop. The form with the stem that returns to the baseline may be more suitable when you want to join a subsequent letter like ‘e’ or ‘i’.
The “stemless d” is not exclusive to Kurrent or Fraktur. It can also be spotted in italics like Cochin’s, among others, and even in Antiqua or Grotesk. It is quite common in German (roundhand) script typefaces from the 1950s, like Paul Zimmermann’s Impuls, Heinrich Pauser’s Petra (1954), G.G. Lange’s Boulevard (1955), Georg Trump’s Time Script (1956), or Helmut Matheis’s Verona (1958). Older examples with this form include Walter Höhnisch’s Skizze (1935), Erich Mollowitz’s Forelle (1936), and Carlos Winkow’s Gong (1945).