My real problem is that I don’t have a usable list of Cyrillic kern pairs so I don’t know what constitutes an appropriate pair in real use.
That depends on the range of Cyrillic you are covering. There are indeed some combinations that never occur (either phonetically impossible or each letter belongs to different languages), but I’d say it’s safer and faster to kern everything than to do research and avoid unnecessary ones. And don’t forget to kern against punctuations too, such as «…» „…“ which are used for quotation in Russian and other countries.
Thanks, Tosche and Jakob; that’s good information, probably exactly what I need – especially the links.
Maybe it would be a good feature to have a Glyphs kern pair file which could be enabled to display kern pairs automatically so the kerning could be set simply by selecting any pair and making the visual adjustment as is done now by typing or pasting.
It would only display a list of kern pairs where both glyphs of a pair exist in the font.
It would work in the same manner the GlyphData.xml file currently does, with a user file overriding the furnished file.
Such a feature would be useful to users worldwide who need kerning help with an unfamiliar language. It would save time in that the user wouldn’t have to set useless kerns because of that unfamiliarity.
I was about to suggest the Cyrillic version of the testing page, but Jakob beat me
I think sample string is just what you want, isn’t it? I customise it quite a lot for kerning all kinds of scripts. Another thing is that the pairs you need to kern differ greatly from one design to another, so such an internal library would end up listing all possible pairs. The most economical way would be that a user builds pair lists for each project, which is what I do in sample string, or in Pair List Builder in Metrics Machine (Metrics Machine is a dedicated kerning app, yet it doesn’t have such list; it assumes you build it each time, which is quite understandable).
At least there are pairs made of letters that belong to different languages (e.g. ЂҠ is a pair of Serbian and Bashkir letters, so it never happens, as of today, in theory). This level of pair elimination may be reasonable. But even if you think kerning between Latin and Cyrillic letters are totally useless, there are languages which mix them (e.g. Wakhi), which are mind-blowing.
Tosche:Sometimes, but not always. In my experience, kern pairs work just fine. I always plan my basic spacing before I ever start a font and any kerning done at the end is just fine-tuning individual cases. Typically I use Kerning Groups for efficiency, and my Cyrillic characters are included, but there are always exceptions. That is correct since it is the Master List. However, it would only display in Glyphs those characters which actually exist in the current font. Exactly, and that is what I am proposing with my idea; we may just be stating it differently. I understand that such things do occur, but those combinations can adequately be covered by using Kerning Groups, I think. It all depends upon the font and the language, really.
I wasn’t asking for a comprehensive kern pair list covering all scripts, only Cyrillic.I agree, and I certainly wouldn't ask Georg to devote his resources to such an enormous task just to benefit a small group of users. But if he could enable users to provide their own lists, that is something he might conceivably be able to do.
I’m sorry, I did not remembered is correctly. It only splits if it has to flatten kerning (because of exemptions).
But I use the same classes for all scripts, anyway.
Does anyone know for sure if it is a problem or what are the potential problems with cross script classes?
Because of the way I build fonts, I can’t see that it would be much of a problem for me or my users other than having kerning pairs in a font that may never occur in use. So far as those extra kern pairs causing excessive processing time, I would like to see a benchmark test of that before I felt it is a problem.
My understanding is this. There are four groups of pairs if you kern between Latin and Cyrillic:
Typically the groups 2 and 3 are never accessed, yet the font renderer actively looks at them even if there is no pair occurring in the current text. It gets even worse when you add Greek to this.
I am just about to start kerning a Latin-Cyrillic-Greek font and I am terrified about the idea of having 9 groups of pairs!!! Does anyone know whether Glyphs finally manages to break the kerning groups by script, thus avoiding this subsequent processing issue?
When I tested this, splitting the classes by script automatically produced bigger GPOS tables. And if you keep the number of pairs at a sane number you should be save. All my fonts are done like this and I never had any problem.
Thanks for the info, Georg.