I’m preety new to the whole subject, so please excuse my question … I created a font which is quite big. language support from greenland to russia, full set of smart caps, tons of ligatures and of course, i get the
GPOS feature ‘kern’ causes overflow of offset to a sub
table
message while exporting. is there any possibility to fix something or do i have to cut the font into maybe three with a regular, a small caps and a ligatures set? the kerning/metrics are from igino marini and it would be a pity to loose that quality …
The one thing you can do is to use kerning classes as much as possible, even if the class ends up with only one member. If you add classes after you added the kerning, you need to run the compress command from the gear button in the kerning window.
How many kerning pairs do you have?
In Glyphs 2 there is one option to activate ‘Extension Kerning’. That saves the kerning in a different way that might not work in all apps (my testing found no problems so far).
Technical background: Each pair in the “kern” table takes 6 bytes, there is a 14-byte header, and the maximum size of a subtable is 64K bytes, so (65536-14)/6 gives you 10920. Most font engines only support a “kern” table with one subtable, so there’s the limit.
The number of kerning pairs gives only an estimate it the font will export or not. Because if you have a pair between a class and a single letter, there will be a pair for each member of the class. Thats why I suggested to add a group to each glyph.
Glyphs does not export a kern table. It exports a kern feature in the GPOS table. The biggest difference is that the kern feature support classes that can reduce the number of needed pairs.
will do that first. thank you. another topic hindering the workflow with iKern was the naming of the cyrillic set because »-« is »vorbidden« so i named them afiiXXXXX what worked out well. may i rename them again into nice names or is that a case for the otf police?