Re: Bootstrap GCC Pass 1 or 2? (was Re: Resolution)
Greg Schafer gschafer at zip.com.au writes: The bottom line is we still no don't know the cause of the issue you are seeing. Until we understand all the issues, I'm very reluctant to majorly alter a build method which has held us in good stead for approx' 4 years. This problem is so far confined to hosts with 64-bit kernel running mostly 32-bit userland and it's possible the only sane solution for this scenario is cross compilation. It might be worth trying a different host distro, Fedora maybe, to see how pervasive the problem is. Anyhoo, I'll try to reproduce and figure out the problem when I get time, but it won't be for a while yet. I'll see if I can try another distro, like Fedora, as you say. To summarize, the tests so far seem to show that the problem is related to the build method producing a binutils that is incompatible with the host: * Using '-B /usr/bin/' for GCC makes the bootstrap fail at stage3. * Dropping the '-B /usr/bin/' argument for GCC makes the bootstrap fail at stage2. In short, the stage that is assembled and linked by the binutils in /tools cannot build another stage. -- JH -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
Re: stray references
El Lunes, 27 de Agosto de 2007 14:17, [EMAIL PROTECTED] escribió: SVN-20070820 When I compiled PCRE, and others, butterfly-build appears in the compiler output and harmful or not, that is unclean and unacceptable. On a LFS or CLFS-based system, both current and old versions, there is a lot of references to /sources/gcc-build. That is due /usr/lib/libstdc++.la and /usr/lib/libsupc++.la contains: # Libraries that this one depends upon. dependency_libs=' -L/sources/gcc-build/i686-pc-linux-gnu/libstdc++-v3/src -L/sources/gcc-build/i686-pc-linux-gnu/libstdc++-v3/src/.libs -lm -lm -lm -L/sources/gcc-build/gcc -lgcc_s -lc -lgcc_s -lm -lgcc_s -lc -lgcc_s' /usr/lib/libbfd.la contains also references to /sources/binutils-build Rebuilding GCC and Binutils don't solve that, the sources build directory is always referenced on its .la files. Thus, I see three possible solutions: - To patch GCC and Binutils to not include references to the build tree and to remove all that diplicated -lgcc_s -lm - To edit manually that .la files. - To not install GCC and Binutils .la files, like most distros do. -- Manuel Canales Esparcia Usuario de LFS nº2886: http://www.linuxfromscratch.org LFS en castellano: http://www.escomposlinux.org/lfs-es http://www.lfs-es.info TLDP-ES: http://es.tldp.org -- http://linuxfromscratch.org/mailman/listinfo/hlfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
Re: stray references
El Lunes, 27 de Agosto de 2007 19:46, M.Canales.es escribió: Actually, the solution for the book is do nothing, IMHO. That references to the GCC and Binutils build trees has been on all *LFS-based systems from years ago without known issues. -- Manuel Canales Esparcia Usuario de LFS nº2886: http://www.linuxfromscratch.org LFS en castellano: http://www.escomposlinux.org/lfs-es http://www.lfs-es.info TLDP-ES: http://es.tldp.org -- http://linuxfromscratch.org/mailman/listinfo/hlfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
Ticket #2066 - automake-1.9.6 fails gcj4.tests
In http://wiki.linuxfromscratch.org/lfs/ticket/2066, the submitter says: automake-1.9.6, when I get to the make check step, it fails one test. Several letters back and forth with the Ralf @ automake, revealed two things: 1: gcj is never installed by the LFS process. 2: automake's gcj4.test didn't check for this. 1. Is this valid for automake-1.10? 2. Should a sentence be added to the book to point out the failure? 3. Does this rise to an issue for the 6.3 release? gcj is the gcc Java compiler. One fix is to build gcc --enable-languages=c,c++,java in Chapter 6, but I don't think this should be done in LFS for just this. -- Bruce -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
Re: LFS 6.3-rc2 acknowledgments page
On 8/12/07, Jeremy Huntwork [EMAIL PROTECTED] wrote: On Sat, Aug 11, 2007 at 08:19:28PM -0500, Bruce Dubbs wrote: I've redone the acknowledgments page in my sandbox, but have not committed yet. Does this look OK to everybody? http://www.linuxfromscratch.org/~bdubbs/lfs-book/appendices/acknowledgements.html I'm no longer the ALFS project leader. But, for that matter, I don't know who is. Development on jhalfs continues via Manuel and George B. (who, btw, isn't mentioned), but not sure if that's necessary for the actual LFS book. In fact the actual ALFS project is a little unorganized, I think. BTW, you mention some CLFS developers, but not all. And some BLFS devs, but not all. What is the prereq. for getting mentioned? Lastly, Ken Moffat is missing. He has LFS commit privs, I believe... Yep, his name even appears in the Changelog. For now, I'm just going to add Ken to the acknowledgments and remove ALFS project leader from Jeremy's description. If anyone feels this should be updated further, please speak up. Otherwise, I'd like to punt further changes to post-6.3. -- Dan -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
Re: LFS-6.3-rc2 has been generated
On 8/13/07, Bruce Dubbs [EMAIL PROTECTED] wrote: ftp://anduin.linuxfromscratch.org/LFS/lfs-packages/version;/shadow-shadow-version;.tar.bz2 http://anduin.linuxfromscratch.org/sources/LFS/lfs-packages/version;/shadow-shadow-version;.tar.bz2 These will be symlinks to conglomeration and all the packages are there. I'm going to push these to trunk now and merge it back to 6.3. -- Dan -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
Re: Ticket #2066 - automake-1.9.6 fails gcj4.tests
On 8/27/07, Bruce Dubbs [EMAIL PROTECTED] wrote: In http://wiki.linuxfromscratch.org/lfs/ticket/2066, the submitter says: automake-1.9.6, when I get to the make check step, it fails one test. Several letters back and forth with the Ralf @ automake, revealed two things: 1: gcj is never installed by the LFS process. 2: automake's gcj4.test didn't check for this. 1. Is this valid for automake-1.10? I don't know. 2. Should a sentence be added to the book to point out the failure? If 1. is yes, then I think so. 3. Does this rise to an issue for the 6.3 release? Well, if it's one sentence to warn for a failure, then I think that should be added. gcj is the gcc Java compiler. One fix is to build gcc --enable-languages=c,c++,java in Chapter 6, but I don't think this should be done in LFS for just this. Obviously, I don't think this is a good idea for 6.3 since it's a little late in the game to be adding another compiler to the game. I also don't think it would be wise to add gcj here. For one, having a java compiler at all is something of a niche area considering the scope of the *LFS books where practically everything is C/C++/Perl/Python. Also, this would create unnecessary conflicts with BLFS, where there are two possibilities for java compilers in gcj and jdk. Just not worth the effort, IMO. The patch in automake will make the tests do the right thing for the next release. It should be noted that automake will try to use tons of different tools that we don't install in LFS, so gcj would not be exceptional. -- Dan -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
New bash/readline patches for 6.3?
I'm checking for opinions on including the new bash and readline patches from upstream. http://wiki.linuxfromscratch.org/lfs/ticket/2067 http://wiki.linuxfromscratch.org/lfs/ticket/2068 I would personally like to include these as upstream provided patches for bug fixes are usually safe. I completely understand if people would rather push the release now. If people are interested in having it for 6.3, I can generate the patch and test it out. -- Dan -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
Re: New bash/readline patches for 6.3?
On Mon, 27 Aug 2007 15:08:01 -0700, Dan Nicholson [EMAIL PROTECTED] wrote: I'm checking for opinions on including the new bash and readline patches from upstream. http://wiki.linuxfromscratch.org/lfs/ticket/2067 http://wiki.linuxfromscratch.org/lfs/ticket/2068 I would personally like to include these as upstream provided patches for bug fixes are usually safe. I completely understand if people would rather push the release now. If people are interested in having it for 6.3, I can generate the patch and test it out. By coincidence, I actually generated the patches (at http://www.linuxfromscratch.org/~matthew/patches/) this evening, but haven't had a chance to even compile test them yet. There's also updated upstream patches for Vim, which I guess should go in if the Bash and Readline patches go in. Whilst I'd prefer to see the updated patches in the book, we're going to have to cut the release at some point, and the longer we leave a hard freeze, the higher the chances that these patches will require further updates! Regards, Matt. -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
Re: Ticket #2066 - automake-1.9.6 fails gcj4.tests
On 8/27/07, Dan Nicholson [EMAIL PROTECTED] wrote: On 8/27/07, Bruce Dubbs [EMAIL PROTECTED] wrote: In http://wiki.linuxfromscratch.org/lfs/ticket/2066, the submitter says: automake-1.9.6, when I get to the make check step, it fails one test. Several letters back and forth with the Ralf @ automake, revealed two things: 1: gcj is never installed by the LFS process. 2: automake's gcj4.test didn't check for this. 1. Is this valid for automake-1.10? I don't know. It seems that 1.10 does the right thing by checking for gcj and skipping the gcj4.test. The tests were clean for me. So, I think this only affects the old version. 2. Should a sentence be added to the book to point out the failure? If 1. is yes, then I think so. I think we're OK. Someone could add a note to the 6.2 errata if they want. -- Dan -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
Ready for 6.3?
As far as I know, all outstanding issues for 6.3 are complete. I'd prefer not to do the new bash patches quite yet. The changes from -rc2 are: man-db fixes patch Linux-2.6.22.5 shadow URL minor bash testsuite fix LFS-Bootscripts-20070813 acknowledgements page If we don't get this out soon, we will continue to have new packages that would be nice to add. I'd rather release and get on with 7.0. Are there any objections to releasing what is in branches/6.3 tomorrow night as 6.3 final? -- Bruce -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
Re: Ready for 6.3?
On Mon, Aug 27, 2007 at 10:28:30PM -0500, Bruce Dubbs wrote: Are there any objections to releasing what is in branches/6.3 tomorrow night as 6.3 final? We have to let go at some point. Now seems a decent enough time. -- JH -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page