On 2021/01/12 06:04, Joel Knight wrote: > On Tue, Jan 12, 2021 at 4:22 AM Stuart Henderson <s...@spacehopper.org> wrote: > > > > I don't get multicast from my ISP any more so can't test runtime. > > ports-wise the hardcoded CFLAGS should go and plist regenerated, > > otherwise looks good. I've added DEBUG_PACKAGES while there. > > Oops, thanks for catching the plist update. I brought CFLAGS over > because it was in the existing patch-Makefile, but maybe the way I did > it should've been conditional? Having -Wall was nice because it caught > some fallout while updating the patches. I couldn't find any guidance > for things like CFLAGS in the porting docs.
The old patch made it so CFLAGS from the system/user was honoured, the new one forces -O. I'll tweak it to add -Wall but opt levels are meant to be under ports control. > > Anyone else want to test or should I just commit it? > > I've been running it myself just between two local VLANs without any > regression compared to v0.1-beta2 in ports. > > > It is probably not worth arguing with upstream about cdefs ;) > > Yeah, I saw where it was headed and backed away slowly :)