From mboxrd@z Thu Jan 1 00:00:00 1970 From: "John David Anglin" To: dave@hiauly1.hia.nrc.ca (John David Anglin) Cc: jason@cygnus.com, law@cygnus.com, gcc-patches@gcc.gnu.org, egcs-bugs@egcs.cygnus.com Subject: Re: g++-mike-eh8-C test failure Date: Fri, 28 Apr 2000 15:41:00 -0000 Message-id: <200004282241.SAA15346@hiauly1.hia.nrc.ca> References: <200004282031.QAA15035@hiauly1.hia.nrc.ca> X-SW-Source: 2000-04/msg01232.html > Changed my mind. The problem appears to be caused by the selection of > tinfo2.o in libgcc.a to resolve "char type_info node" (__tic). After much testing, it appears that the hpux linker prefers to resolve DATA symbols from archive libraries even when a shared library with the symbol preceeds the archive library in the command. It only seems to resolve DATA symbols from shared libraries when all symbols are explicitly exported with "-E" or when no archive library contains the symbol. Yuck! It looks like the global DATA symbols that reside in both libstdc++.sl and libgcc.a should only be in one or the other (probably libgcc.a). -- J. David Anglin dave.anglin@nrc.ca National Research Council of Canada (613) 990-0752 (FAX: 952-6605)