[feaLib] Correct ChainContextPosStatement and ChainContextSubstStatement #3788
Conversation
They don’t have self.glyph.
“Liang” is my surname.
|
Since this code is not covered by any tests, can you add some? |
|
Uh, I don’t really understand the intended behavior of this part of code (this PR just fixed some apparent typos), and https://github.com/fonttools/fonttools/tree/main/Tests/feaLib seems rather complicated. I’ll leave the test case to you guys… |
|
I thought you hit this with some feature code, but it is alright if you don’t have a test. I’m also not sure about this special case. |
|
Oh I hit this when I generated FEA code from the AST. I accidentally created a ChainContextSubstStatement instance with “no prefix, no suffix, and no nested lookups”. Parsing from a sub rule without the |
Yes, I guessed it would be something like this. Too niche for a test. |
They don’t really have
self.glyphs. The erroneousself.glyphhas been hiding in a branch of edge case (no prefix, no suffix, and no nested lookups).It’s unclear if such an edge case should be serialized as a special case (without the
'mark), therefore I’m only proposing a minimal fix for now.