Macro expansion difference vs other preprocessors

[ehaas]

Found this example here: https://marc.info/?l=boost&m=118835769257658&w=2

#define A(m) m( B(f)
#define B(x) A(x)

#define C(x) < x >

A(C) ) *

clang, GCC, MSVC, and tinycc output:

< f( B(f) > *

arocc output:

<  A(f)  > *

The post says that our output is correct; and that "Very very few preprocessors get this completely right". Not sure if this is a bug or not.

[Vexu]

Assuming that the post and our expansion is correct it could be argued that this is a bug since we currently always emulate some other compiler and our behavior doesn't match. Did your preprocessor rewrite get far enough that you could test how it handles this?

[ehaas]

Unfortunately not; I abandoned it because I wasn't able to find big enough chunks of time to work on it, so every time I came back to it I'd have to re-remember how preprocessors work. I might be able to take another try toward the end of the year.