Joe Orton wrote: > On Wed, Jun 01, 2005 at 05:09:59PM -0700, Philippe M. Chiasson wrote: >>Ian Holsman wrote: >>>Philippe M. Chiasson wrote: >>>>Ian Holsman wrote: >>>> >>>>>I just tried this on a EL3 machine, and it goes further (ie.. it can >>>>>start apache) >>>>> >>>>>the version of perl is the same, but the httpd on EL3 is 2.0.52 and the >>>>>GCC version is 3.2.3 (EL4 has 3.4.3) >>>>> >>>>>can anyone else reproduce this? >>>> >>>>Sounds like this problem is hitting you too: >>>> >>>>http://marc.theaimsgroup.com/?l=apache-modperl&m=111686159013437&w=2 >>>> >>>yeah.. I was just about to post that getting a non-LFS build of httpd >>>fixed the issue ;-) >> >>Cool, just happened to be fresh in my memory ;-) >> >>Would be nice to figure out why we don't detect this particular condition >>and at least abort at perl Makefile.PL time. Last I check, we were supposed >>to catch these kind of mismatches. > > There is no "conflict" to detect in this situation, really; both Perl > and httpd are built with LFS support, so in theory it should work. > > But 2.0.x does not support LFS, people who build with > -D_FILE_OFFSET_BITS=64 are playing with fire and should completely > expect everything to burn down around them at any moment. This > combination is untested and unsupported, and it is known to break in > horrible ways with various third-party modules.
Yes, and as can be seen by that original segfault, you can't expect a combination like this to work. Hoever, If it's a problem likely to appear to many of our users, I'd rather at least try and warn them about it. "Whoa there, you are trying largefiles with httpd-2.0.x, enjoy the segfaults!" -------------------------------------------------------------------------------- Philippe M. Chiasson m/gozer\@(apache|cpan|ectoplasm)\.org/ GPG KeyID : 88C3A5A5 http://gozer.ectoplasm.org/ F9BF E0C2 480E 7680 1AE5 3631 CB32 A107 88C3A5A5
signature.asc
Description: OpenPGP digital signature
