I agree, ``conn'' is probably not the best choice, especially inside a commonly included header file.
Interestingly enough, I was trying to think of reasons this compiled fine for a friend of mine, but not me. Since this is a new OpenSolaris install, GNU make wasn't around, so typing make invoked Sun's make. Using gmake made everything compile fine. I haven't really looked into why this is, but it worked. Joel On Fri, Jan 9, 2009 at 11:36 PM, Dale Ghent <[email protected]> wrote: > On Jan 9, 2009, at 12:43 PM, Joel wrote: > >> >> afs_nfsdisp.o -c /space/openafs/openafs-1.4.8/src/afs/afs_nfsdis >> p.c >> "/space/openafs/openafs-1.4.8/src/afs/afs.h", line 294: (struct) tag >> redeclared: conn >> >> From what I can tell, conn is defined in /usr/include/rpc/rpc_rdma.h >> which is included in afs_nfsdisp.c, AND in afs.h. Both places are >> important, so renaming is probably out of the question. Any thoughts? > > This is going to be a pain in the ass to fix or work around. > > On one hand, shame on the NFS RDMA team for instituting such a > generically-named global. On the other, shame on openafs for doing the same. > But openafs was there first. Oh well. > > Looking at the code, there doesn't seem to be a way we can #ifdef our way > around this. If might be easier for us to just rename struct conn to > something more specific, such as afs_pconn. > > O elders, what sayeth thou? > > /dale > _______________________________________________ OpenAFS-devel mailing list [email protected] https://lists.openafs.org/mailman/listinfo/openafs-devel
