On Sun, Jan 11, 2009 at 1:04 PM, Matt Benjamin <[email protected]> wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA256 > > Hi, > > Looks like this issue is also RT 119118, filed Oct. 3. I support the > suggestion. However, wouldn't the preferred name be "struct afs_conn"? > > With regard to the gmake story, I'm suspicious. I'm told gmake is the
I'm also quite suspicious. The libafs build system is rather different from the rest of OpenAFS -- src/libafs/Makefile uses shell for loops to call down into build-time created subdirectories in order to build the 32- and 64-bit kernel modules. My recollection is that /usr/ccs/bin/make and gmake handle error code propagation from the shell for loop-invoked submakes differently. As a result, it's possible one make variant is silently failing, while the other is dropping dead in the water immediately. So, to repeat Matt's question: are you sure it actually built libafs.o with gmake, or did it just fail to abort on submake failure? -Tom _______________________________________________ OpenAFS-devel mailing list [email protected] https://lists.openafs.org/mailman/listinfo/openafs-devel
