Compress kerning & Compress again

I have a MM font with kerning classes in GLyphs. The kerning of each individual master is produced in an external app, in FL. In FL I have around 700 kerning pairs. Then kerning is expanded, each master is exported as individual UFO, imported to Glyphs and from there copy-pasted into the MM file into the corresponding master.

So now I have 4 masters with around 15.000 kerning pairs. I press compress to adapt my kerning pairs to my GLyphs kerning classes. After I press compress once I get around 4000 pairs and kerning (seems to) work ok. If I press compress again I get around 800 but the kerning doesn t work correctly and I have the feeling that kerning classes and spacing classes get messed up.

For example for ‘k’ glyph I use ‘n’ as a metric class but ‘h’ as a kerning class. So after I press compress for the second time all the glyphs contained in the ‘h’ kerning class get the same kerning as glyphs in ‘n’ metric class so for example I get an overlap in the combo ‘Th’ while ‘Tn’ works fine.

I do not know if I m doing a wrong methodology to import external kerning, if I m doing something wrong or if I am reporting a bug. What does that second compress step do??

There is no such thing as a spacing class in Glyphs, only metric keys. It would make more sense to export the FontLab file directly to a .glyphs file, update the glyph info, and then copy and paste the kerning data.

ok but what about that second compress step? why does it change the kerning?

I’d have to see the font for that. I suppose something got lost in the UFO conversion. Please try exporting directly to .glyphs. If the problem remains, send the .glyphs file to support at this domain.