https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124009

--- Comment #4 from Jeffrey A. Law <law at gcc dot gnu.org> ---
Tackling spaceship is definitely on the todo list, but I think there's still a
reasonable chance this is going to show up in practice even if we implemented a
spaceship expander.  Perhaps the biggest advantage of spaceship in this context
would be the simpifications show up earlier and thus we're a better chance
they're participate in CSE, fwprop, etc.

It's also the case there are meaningful generalizations.  Essentially we just
need to implement the select between 2^n and 0 and the 2^n-1 and -1 case is
trivially synthesized by existing code.

Reply via email to