(Illustration by Gaich Muramatsu)
On Fri, Feb 01, 2002 at 05:21:06PM -0500, Victor wrote: > > Now that you mention it, the link line doesn't include -lrpc2 because > > dirtest shouldn't even be using any functions from 'getsecret', so why > > is the linker complaining? > > Compiler verson: > $ gcc -v > Reading specs from /usr/lib/gcc-lib/i386-redhat-linux/2.96/specs > gcc version 2.96 20000731 (Red Hat Linux 7.1 2.96-98) > > Damn, is it because of 2.96? Could very well be, I've used 2.7.* and 2.95.* successfully. We've already managed to identify one case where 3.0.* was miscompiling Coda. And I believe 2.96 was a 3.0 beta version, so it can very well have not just the normal 'Coda is doing something stupid with C vs. C++', but also the 'beta-3.0' bugs. > I added it and dir compiled fine. But then another error came right > afterwards: > ... > gcc -fno-exceptions -L/usr/coda/lib -L/usr/coda/lib -L/usr/coda/lib > pdbtool.o libal.a ../util/libutil.a /usr/home/victord/CO > DA/coda-5.3.17/lib-src/base/libbase.a -ldb1 -lreadline -ltermcap -o > pdbtool > ../util/libutil.a(getsecret.o): In function `GenerateSecret(unsigned char > *)': > getsecret.o(.text+0x1f2): undefined reference to `rpc2_NextRandom' Same problem, pretty much everything linked against libbase will have the same thing. And neither splitting up libbase, nor linking everything against librpc2 is a good solution. Maybe not using the rpc2 random number generating function in lib-src/base/getsecret.c would be better? JanReceived on 2002-02-01 18:05:42