I know we aren't using git, but I would like to see git-style commits in cvs, where each 'cvs commit' invocation is a single logical change. That makes it far easier to understand history. I don't mean that each patch should be split, but fixing upstream bugs and changing makefiles to add clang support seem like separate things. The patches are not all obvious. I don't particularly suspect that any of them are incorrect, and I think odds are very good that they are all ok. But they do not have comments, and there are not links to the upstream bugtracker. It seems like many of them should be upstream. I am bringing up the comments/upstream-tracker-links issue because I am not really comfortable with pkgsrc being used to indefinitely carry patches that belong upstream. While it's progress to have the patch, it also means that anyone doing updates has a lot more work to do. Also, I don't think this hunk should be applied: +- if (write(fd, &out, sizeof(uint32_t))) ; ++ if (write(fd, &out, sizeof(uint32_t))) ++ ; That's a style violation (probably; I haven't read upstream's style guide), but it's not a bug, This is the sort of thing that should be reported upstream and we will eventually get the fix. If they don't want to change it, then there's no reason to have a patch in pkgsrc which is just cognitive load on updaters. You didn't mention testing; have you booted the various xen kernels (compiled with system gcc, at least) on some combination of netbsd-5 and netbsd-6 and seen that dom0 and a domU are ok? I know we don't have a testing policy, but xen stability seems critical and I'd like to hear what the people who tend to deal with xen think.
Attachment:
pgpjXjiFcSnpa.pgp
Description: PGP signature