no_new_pseudos
Richard Kenner
kenner@vlsi1.ultra.nyu.edu
Thu Jul 5 02:07:00 GMT 2007
> Alexandre> It's as mechanical as the change you proposed, except that yours
> Alexandre> potentially loses information that would enable someone to recover
> Alexandre> !BEFORE_RELOAD_P() out of the expanded version of no_new_pseudos.
>
> Except that no_new_pseudos was not used consistently.
That's my concern too. As unpleasant as it might be, I think each
occurrence has to be looked at individually. Given that many uses
might not be correct, I doubt there's any shortcut here.
However, I do like the idea of those predicates once this has been sorted out
because I think it'll help avoid this sort of problem in the future.
More information about the Gcc
mailing list