On Mon 02 Oct 2023 at 18:33:13 +0200, Rhialto wrote: > On Mon 02 Oct 2023 at 16:19:19 +0200, Benny Siegert wrote: > > Is this repeatable? That is, if you just re-do "make", does it fail at > > the same place? > > > > I have seen segfaults from gcc before, typically on flaky hardware. > > They do not reappear when rerunning the same command. > > Yes, it's repeatable. But it's not gcc that fails, but some other > command: > > [823/1781] Generating gtk/Gdk-4.0.gir ... command (wrapped by meson to set env) > FAILED: gtk/Gdk-4.0.gir (the next line was rather long so I omit it here) If I rerun that command (it runs g-ir-scanner) it outputs ld: warning: libintl.so.1, needed by /tmp/pkgsrc/x11/gtk4/default/gtk-4.12.1/output/gtk/libgtk-4.so, may conflict with libintl.so.8 ld: warning: libintl.so.1, needed by /tmp/pkgsrc/x11/gtk4/default/gtk-4.12.1/output/gtk/libgtk-4.so, may conflict with libintl.so.8 Command '['/tmp/pkgsrc/x11/gtk4/default/gtk-4.12.1/output/tmp-introspectd2c9xgmi/Gdk-4.0', '--introspect-dump=/tmp/pkgsrc/x11/gtk4/default/gtk-4.12.1/output/tmp-introspectd2c9xgmi/functions.txt,/tmp/pkgsrc/x11/gtk4/default/gtk-4.12.1/output/tmp-introspectd2c9xgmi/dump.xml']' died with <Signals.SIGSEGV: 11>. It may well be the conflicting libintl that break things here... libintl.so.1 is from /usr/lib/ and is indeed required by the to-be-built library libintl.so.8 is from /usr/pkg/lib/ but what pulls it into that context is unclear yet. /usr/pkg/bin/g-ir-scanner is a python script and python3.10 also links to the base -lintl.1. Like mixing c++ runtime libraries is bad (and hence having to use the same gcc version for everything), perhaps I should simply force the use of pkgsrc libintl for everything? -Olaf. -- ___ Olaf 'Rhialto' Seibert <rhialto/at/falu.nl> \X/ There is no AI. There is just someone else's work. --I. Rose
Attachment:
signature.asc
Description: PGP signature