Possible error in propagation of anchors

Here's my understanding of the situation:

When composite glyphs like "edieresis" inherit anchors from their components, glyphsLib often has to choose between two anchors with the same name--e.g. "top" on "e" and "top" on "dieresiscomb." In the case of "top" it always (correctly) chooses the "top" anchor from the mark and records that in the glif file.

But I have added an anchor of my own for "usabovecomb" (uni1DD2), which should be positioned differently from other diacriticsI have a "us" anchor on both base glyphs and marks (i.e. for the mkmk feature).

The trouble is, when glyphsLib generates UFOs, the "us" anchor that ends up in the glifs for glyphs like "edieresis" is the one on the base glyph rather than the one on the mark (example abbreviated):

```

...








...

```

The "us" anchor is the one from the base glyph; the one from the mark would have a different x value.

Please excuse if this isn't a glyphsLib problem (I've found similar but not identical issues discussed here) or if I'm missing something obvious.

Start by reproducing the glyphsLib conversion for a composite such as edieresis, using the us anchors on e and dieresiscomb, and inspect the generated GLIF output. Compare the selected us anchor with the mark's expected position; done means the generated UFO uses the intended mark anchor while preserving the existing anchor behavior.

Written by the indexing model from the issue text.