I am resurrecting this thread to see if there has been any movement on this. Currently, I am developing on ancient 5.8.5 && 5.8.9 that explicitly have this problem. The easiest somewhat self contained test where this manifests is in MooseX::Types::Structured's t/06-api.t (since it exercises equals()).
And it isn't limited to ia64 as my environment is x86_64 so it seems to be an issue with 64bit in general. Indeed, forcing numification seems to correct the issue. On Fri, 9 Sep 2011 15:32:19 +0200 Alessandro Ghedini <al3x...@gmail.com> wrote: > On Wed, Sep 07, 2011 at 09:02:01PM -0500, Jesse Luehrs wrote: > > On Wed, Sep 07, 2011 at 08:51:37PM -0500, Dave Rolsky wrote: > > >> * Work around a numification problem on ia64 > > >> > > >> This basically fixes a test failure on the ia64 (Itanium) > > >> architecture. This has been already reported as RT#59478 [1] and > > >> perl #77456 [2], and seems to be already fixed in perl itself > > >> (commit a42d02426c, from v5.13.8). > > > > > > I think doy wanted to come up with a less hacky fix for this. I > > > don't know where he is on this. > > > > What I'd like to do is find a fix that would let plain == work > > anywhere, since there are other places (specifically MooseX::Types, > > but possibly other things too) that also need to know to do > > comparisons this way, and that's less than ideal. > > Are you already working on this (or you already know how to do this)? > If not may I suggest applying the patch for now and adding a warning > note in the docs, until a proper fix is made? > > > The issue is that we don't have any machines available that trigger > > this error, so figuring out how exactly to do this is difficult. If > > anyone has any machines available to do some testing on, this would > > be helpful. > > Debian has a ia64 porter box, but I have no access to it since I'm not > a Debian Developer yet. If you have a patch or something to test, > maybe I can ask someone from the Perl Group who has DD rights to try > it out. > > Cheers >