> I did ask Jeff for these patches. I know it clutters up the source but it is > relatively small and very easy to rip out at merge time. Btrfs gains much > more from broadening the testing base than it does from keeping a few bits > pretty.
We had some slightly more rigorous infrastructure for doing this in the RDS tree before it went over to OFED and had its hands tied by their crazy build infrastructure. The intent was to always maintain the source against current mainline and provide compat headers for older kernels transparently. That worked a surprising amount of the time. When it didn't we chose the ifdefs carefully so that they could be ripped out with a script when we generated a patch for mainline merging. It was what you'd expect.. some autoconf, some --include of little API header files, some awk to weed out the ifdefs, etc. http://oss.oracle.com/projects/rds/src/trunk/kfeature.m4 http://oss.oracle.com/projects/rds/src/trunk/configure.in http://oss.oracle.com/projects/rds/src/trunk/kapi-compat/ http://oss.oracle.com/projects/rds/src/trunk/scripts/makepatch.sh As far as this kind of thing goes, I was pretty happy with it. It got the job done without butchering the code or adding tons of infrastructure. If someone was excited about picking it up for btrfs I'd be happy to answer questions about it, etc. - z _______________________________________________ Btrfs-devel mailing list [email protected] http://oss.oracle.com/mailman/listinfo/btrfs-devel
