From mboxrd@z Thu Jan 1 00:00:00 1970 From: "John David Anglin" To: zackw@stanford.edu (Zack Weinberg) Cc: neil@daikokuya.demon.co.uk, dj@redhat.com, gcc-patches@gcc.gnu.org Subject: Re: cpplib: basename () fix Date: Tue, 06 Mar 2001 10:52:00 -0000 Message-id: <200103061852.NAA16419@hiauly1.hia.nrc.ca> References: <20010306103244.X21391@wolery.stanford.edu> X-SW-Source: 2001-03/msg00304.html > On Tue, Mar 06, 2001 at 01:07:12PM -0500, John David Anglin wrote: > > > I just noticed that Zack in "x-files must die: alloca" has revamped the > > > handling of alloca. I want to look carefully at what he has done. See > > > < http://gcc.gnu.org/ml/gcc-patches/2001-03/msg00194.html >. Possibly, we > > > > In looking at this patch, I see that Zack uses AC_TRY_RUN to determine > > the stack direction for C alloca in libiberty_AC_FUNC_C_ALLOCA. The test > > runs at exactly the same spot that I wanted to put the basename functionality > > check. > > The AC_TRY_RUN in the alloca logic has a fallback in the case we're > cross compiling (it postpones the test to run time). In the basename case, we don't need a fallback because we just use the libiberty version when cross compiling (ie., test fails) unless the runtime environment explicitly excludes it from the list of funcs to check. Dave -- J. David Anglin dave.anglin@nrc.ca National Research Council of Canada (613) 990-0752 (FAX: 952-6605)