Dealing with triplets, let’s say ABC, I want B to take it’s kern to C, from A. How can I do that?
Basically, B has a width of zero (but the actual image is aligned properly above A), so, I just want A (actually a class) to all individually kern to C (another class) as if B was not there. So I suppose I need a little more complicated - I need to do something like this:
Let @B kern to @C by (amount @A kerns to @C) minus (amount @A kerns to @B).
Is there a way to do this?
Thanks!
You need to calculate it yourself and add contextual kerning in kern.
Oh damn!
Thanks for helping answer 
Or… is there another way? Can I somehow make a any character behave like a diacritic? Like for example the comma, behave instead like a diacritic? That would solve the positioning issue and keep the input system for the user very simple and intuitive and give the correct visual results. Or, are only official diacritic inputs allowed to do that?
In the end it is all positional Feature code that is produced, be it by kerning, manually written feature code, or with mark positioning anchors (top and _top, etc.). Kerning creates the kern feature. Manually written feature code goes into the feature you specify. Any glyph that is defined as combining/nonspacing mark and has the required anchors will trigger the creation of (a respective entry in) the mark feature.
Thanks @mekkablue. Hopefully this will get sorted with the marking issue on the other thread! I think it’s that the markers are ligatures that may be causing the problem. Perhaps this is an unusual thing to do with ligatures! But potentially very powerful.