On Fri, Oct 15, 2010 at 04:28:40PM -0400, John Baldwin wrote: > On Friday, October 15, 2010 2:50:46 pm Rui Paulo wrote: > > On 15 Oct 2010, at 13:45, John Baldwin wrote: > > > > > On Thursday, October 14, 2010 5:09:58 pm Dimitry Andric wrote: > > >> On 2010-10-14 21:39, John Baldwin wrote: > > >>> On Thursday, October 14, 2010 3:19:19 pm Rui Paulo wrote: > > >> ... > > >>>> Revert r213765. This is required because our build infrastructure > > >>>> uses > > >>>> the host lex instead of the lex built during buildworld. I will MFC > > >>>> the > > >>>> lex changes soon and in a few weeks this I'll commit again r213765. > > >>> Can't you make 'lex' a build-tool to workaround this? > > >> > > >> That will not help for "cd conf/CONF && make kernel", apparently. It > > >> will always use the host lex. > > > > > > Well, yes, but that is always true. build-tools are only used for > > > buildkernel. However, if an 8.x lex cannot build a 9.x kernel, then > > > having > > > lex be a build-tool (or cross-tool, ru@ knows which category better than > > > I) > > > will let a 'make kernel-toolchain' followed by 'make buildkernel' of a 9.x > > > source tree work on an 8.x host. > > > > Yes, but I was told that 'cd conf/CONF && make kernel' is a supported > > configuration (without requiring kernel-toolchain first). > > Nah, just when it happens to work. It's ok to require people to build a new > world to get a new lex in that case. However, for the buildkernel case the > 'buildworld' / 'toolchain' / 'kernel-toolchain' targets should always build > enough tools to let buildkernel work, so if a new lex is required they should > build a new lex. If it does not cost too much to keep make kernel style of build working, then it should be kept. In this case, the only requirement is to merge lex changes to stable branches. This is what I suggested (or requested, depending on how you interpret my mood). The lex backporting is orthogonal to the issue whether the lex shall become build tool.
pgpd3Jy3Q6Ko2.pgp
Description: PGP signature
