From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Parker, Ron" To: Carl Thompson Cc: "'cygwin@sourceware.cygnus.com'" Subject: RE: problems with latest Cygwin 1.1 Date: Mon, 05 Jun 2000 15:35:00 -0000 Message-id: <23912.960244559.0@NO-ID-FOUND.mhonarc.org> X-SW-Source: 2000-06/msg00159.html > Thank you for your response. Going through the list, I did > find a link to > an updated version of gdb that solved the Insight source > loading problem > (will this go into the main distribution soon?). It's being tested and should become part of "latest" soon. > However, I did not find anything that addresses my main > problem of "g++ > -mno-cygwin" does not produce working executables. This does > not seem to be > an exception problem as using "-fno-exceptions" makes no > difference and the > source uses no exceptions. The whole test program I was compiling was > simply "main(){}" . If I compile it with "-g" and use the > working gdb on > it, it says the program is dying before main() in > "_size_of_stack_reserve__" > . Interestingly, if I run "cygcheck" against the executable > it shows that > it is using "cygwin1.dll" even though I compiled with "-mno-cygwin" . > Clearly there is something wrong with the compiler / linker > when C++ and > "-mno-cygwin" are used. I am running on Windows 2000. > Compiling C using > "gcc" works properly. Also, Cygwin b20 worked. Compiling without > "-mno-cygwin" works too. I could refer you back to the same set of messages, but... Using g++ always causes the application to be linked against libm.a, which is a symlink to libcygwin.a. The work around is to use gcc with the -lstdc++ switch, if needed. This will link without libm.a. http://www.delorie.com/archives/browse.cgi?p=cygwin/2000/05/31/18:07:09 contains more information on the subject. Here is a copy of the patch that is referred to, since the archive do not keep attachments. I will also send you the fixed g++.exe if you would like. It is about 29KB.