| >> ptext (sLit "In module") | >> <+> quotes (ppr (is_mod decl_spec)) | >> <+> source_import <> colon | > | > BTW, I should note that it looks like this is the only instance of this | > problem in the whole of ghc | | So let's just fix ghc here. It looks to me like whoever wrote that code was relying | on a non-obvious consequence of the precedence. At a glance, it looks like there is | intended to be a space before the source_import and colon. And that is exactly what | the proposed fixities for <> and <+> will give. I find it hard to believe that many | people would consciously have relied on the older behaviour. If there's a bug in GHC, or a feature request, can someone make a ticket with a standalone reproducible test case, please? I have not been following the twists and turns of this discussion. Simon