https://gcc.gnu.org/bugzilla/show_bug.cgi?id=123212

--- Comment #10 from Filip Kastl <pheeck at gcc dot gnu.org> ---
My plan for this PR is to do this for -O0:

> So how about at -O0/-Og we tell GCC to just try to convert the whole switch
> into a single jump table?  And if that fails GCC would fall back to the
> "horrid codegen" (search tree :)).  That way we avoid both being slow and
> generating counter-intuitive code.

But I'll wait for Stage 1.  This doesn't seem like a Stage 4 material to me.

Reply via email to