On Tue, Aug 10 2021, Simon Peyton Jones
It's hard to tell what is happening without a repro case. Can you share one?
Haven't been able to do that with <10MB of output, I'm afraid ...
You suggested that it might have something to do with using an operator. Does the same thing happen if you replace the operator with an alpha-numeric name?
I've now concluded several things are coming together. As things started working with INLINE [0] instead of INLINE, it's not the symbolic name. First, reading the ghc source code suggests I can only have one -ddinline-check. Correct? Also, I'm guessing that the inlining I didn't see reported by -dinline-check happened inside the simplifier pass inserted by the ConCat plugin. (And hence INLINE [0] moved it out of that pass.) Is it possible that the flag isn't getting propagated there? (Sorry for being vague - if you don't know offhand, it's not worth digging without more info from me.) -- Regards, Mike