Created attachment 412376 [details, diff] xemacs-21.5.34-gcc5.patch Patch described at http://permalink.gmane.org/gmane.emacs.xemacs.patches/14884 fixes it. If I understand correctly, the patch is already upstream. Quick & dirty backport attached.
Problem has been reported upstream which confirms this bug. Solved by xemacs-21.5.34-r4 which uses a patch based on the patch that finally went in upstream. Note: The ebuild has not been verified with gcc5 due to lack of gcc5 environment. It has only been verified not to break with gcc4. Final verification is to be performed when gcc5 is available in testing or if someone with a gcc5 environment can verify the ebuild.
(In reply to Mats Lidell from comment #1) > Problem has been reported upstream which confirms this bug. > > Solved by xemacs-21.5.34-r4 which uses a patch based on the patch that > finally went in upstream. > > Note: The ebuild has not been verified with gcc5 due to lack of gcc5 > environment. It has only been verified not to break with gcc4. Final > verification is to be performed when gcc5 is available in testing or if > someone with a gcc5 environment can verify the ebuild. Hi, sorry, didn't notice your comment for some time. The -r4 seems to be merging fine on my gcc52 ~amd64 box, thanks! The resulting executable can even be run without segfaulting :)
Thank you for verifying that the patch works.