This is the mail archive of the
mailing list for the GCC project.
Re: [lto] make more things go
- From: Tom Tromey <tromey at redhat dot com>
- To: gcc-patches at gcc dot gnu dot org
- Date: Wed, 21 Nov 2007 20:27:41 -0700
- Subject: Re: [lto] make more things go
- References: <20071121210644.GU24671@codesourcery.com>
- Reply-to: tromey at redhat dot com
>>>>> "Nathan" == Nathan Froyd <firstname.lastname@example.org> writes:
Nathan> This patch doesn't produce any spectacular results on testsuite
Nathan> numbers--it may actually regress slightly due to the builtin change--
Nathan> but (with minor tweaks to the SPEC code to properly declare
Nathan> [*] Yes, I know you're not allowed to do this. What do people think is
Nathan> the right thing to do here? Entirely ignore function types for the
Nathan> purposes of merging decls?
IMO, emit a warning, but have a way to disable it. And, regardless of
the warning, do whatever the linker does now. Maybe don't consider
"bad" functions as inline candidates across object files.
My rationale for this is that real programs do this dumb stuff, and
the above seems to yield the least surprising result.
It would be interesting to know whether anything in gcc triggers this
warning. But I suppose we could find out with --combine.