Typz added a comment.

> Also double-checked with Richard Smith and he agrees that the current 
> behavior is preferable. For comma and plus this doesn't seem overly 
> important, but making it:
>   aaaaaaaaaa(bbbbbbbbb + ccccccccccc *
>                          ddddddddd);
> seems really bad to him as this suggests that we are adding both ccccccccccc 
> and ddddddddd.
>   aaaaaaaaaa(bbbbbbbbb + ccccccccccc *
>                              ddddddddd);
> seems clearer.

And I fully agree with this!
But I don't think that would be affected by my patch... I hope it does not at 
least, and I'll try to double check (and add a test).

> And for consistency reasons, we should not treat those two cases differently.

The consistency is related only to the implementation of clang-format: for a 
language (and its users) they are not the same things, comma separating 
arguments are not operators and are used to separate different expressions.

  rC Clang


cfe-commits mailing list

Reply via email to