It appears that Glyphs automatically generates code to use uppercase marks when two combining marks are stacked on top of each other, even when applied to lowercase letters. While this may be appropriate for some use cases (maybe Vietnamese…?), it is not always desirable. Specifically, I think it is inappropriate for linguistic phonetic transcriptions, and I have never seen this done in any fonts commonly used for typesetting IPA. I can work around this by manually removing the two generated lines from ccmp every time I update the features, but is there a more elegant way to disable this?
Besides, the code as it is currently written seems to produce inconsistent results with fonts that include precomposed characters. Since õ exists precomposed in the font, ṍ does not get targeted for uppercase mark substitution, but there is no precomposed ə̃, so ə̃́ gets the unfortunate compressed appearance shown here.
