On Sat, Feb 23, 2013 at 08:13:36PM -0600, Bruce Dubbs wrote: > > It turns out that the above warning does not actually stop the build > like s/item/itemx/ does. > > > Leave it with me - even if a larger patch does solve everything, I > >guess a workaround will still be needed in chapter 5 for building > >from recent hosts (not just Arch, but also 7.3 building itself or > >building next month's svn).
I failed to achhieve that - downloaded the gcc 4.7 svn branch last night, but everything in it is dated when I downloaded, and therefore newer than 4.7.2. Started to compare the info files, but that was going to take hours. Tried diffing the branch to 4.7.2 and editing it down, but looks like that will take _days_. So, many thanks for what is below. > > Attached is the diff I got by hacking the .texi files at each error > and re-running make. I then extracted the source tarball again and > did a diff and edited out the non-applicable files from the actual > build. If you do the same for testing, be sure to remove the mpfr, > mpc, and gmp directories. > > The diff applies to: > > patching file gcc/doc/cppopts.texi > patching file gcc/doc/gcc.texi > patching file gcc/doc/generic.texi > patching file gcc/doc/invoke.texi > patching file gcc/doc/sourcebuild.texi > > There are still a lot of warnings in the .info build, but the > overall gcc build completes. I have not retested tar -xf; patch; > build. > Works for me. Tested with info -f in the build tree and, after DESTDIR install, info -d. Seems to work fine for gcc, g++, cpp, libgomp, libitm, libquadmath, i.e. all the parts we install in LFS. > I still am not in favor of putting this in LFS-7.3. It's so much > easier to omit the .info build completely and, of course, there is > no sense at all in building it in Chapter 5. > > -- Bruce > I guess I have to agree with you - if anyone rebuilds gcc in BLFS to add extra languages then the build will fail in (at least) ada, fortran, gcj unless the sed from LFS is used. I'm not going to try to diff the remaining info files (I can't test ada or gcj) or otherwise hack them. Please consider heading/naming that diff to patches- standards and committing it so that anyone using only C, C++ can include it and be able to look at the current descriptions of the -m and -f options. ĸen -- das eine Mal als Tragödie, das andere Mal als Farce -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page