On 2012-08-27 05:00, Ken Moffat wrote: > On Mon, Aug 27, 2012 at 03:46:00AM +0100, Jasmine Iwanek wrote: >> >> As I've stated several times now, we should *not* be pulling host >> headers into glibc. >> > You haven't explained why you didn't install the headers on youur > first build. In page 6.9 we say > > Install NIS and RPC related headers that are not installed by > default; these are required by several BLFS packages: > > cp -v ../glibc-2.16.0/sunrpc/rpc/*.h /usr/include/rpc > cp -v ../glibc-2.16.0/sunrpc/rpcsvc/*.h /usr/include/rpcsvc > cp -v ../glibc-2.16.0/nis/rpcsvc/*.h /usr/include/rpcsvc > > Sure, we've only started this discussion because you didn't do > that, and some of the results have been interesting, but we _expect_ > those headers to get installed (like they were in past versions of > glibc). > > Do we need to use stronger language, or perhaps avoid the > explanation so that people who think "I don't care about XXX in > BLFS" will not omit them ? I'm curious. >
Maybe I've made an omission on my part too, I'll be a bit clearer if I can: We shouldn't be pulling in headers to the build in ch5 after binutils pass1 and gcc pass1 which are in /usr/include, by that point the compiler and linker *should* be using /tools/include instead. I omitted those instructions from my build initially due to having nothing that needed them, this is what led to me noticing that we have the include problem. What can I say, I'm a purist, I'd left them out after that simply due to to the fact nothing I was adding from BLFS had asked for them. My *initial* post being due to simply having forgotten them, from after that, noticing that it was using /usr/include when it shouldn't. > ĸen > -- > das eine Mal als Tragödie, das andere Mal als Farce -- Jasmine Iwanek -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
