On 1/25/06, Matthew Burgess > > What I'd like to do now is: > > 1) Have the two toolchain bugs (1675 and 1677 - note that 1675's title > isn't entirely accurate!) fixed in LFS trunk. I'd need to re-read the > discussions on those two to figure out quite what's wrong and how to fix > them, or someone else could just post patches and I'll apply them > instead :-)
1. 1675 is good to go with the sed proposed. I've tested it numerous times on a couple x86 boxes. I'll whip up a patch tonight. The wording will need to be worked on. This would also allow us to drop the libidn add-on from Ch. 5 glibc (not much a winner, though). 2. I 100% agree with what Alexander has written for 1677. If you wiped your directory and started again, your unpacked /tools would not work. This hits on the Ch. 5 stuff. Drop --enable-shared from Binutils Pass 2 and we can do the symlinking to /tools/bin/ld-new like DIY does. Go to the -specs=/tmp/myspecs workaround instead of mangling your specs file in /tools. 3. One more thing. I didn't find a bug about this, but Alexander brought up the Ch. 5 toolchain in lfs-dev the other day: http://linuxfromscratch.org/pipermail/lfs-dev/2006-January/055263.html This might be the most important because it is definitely a show stopper. Blacklisting certain RH distros is not the way to go. I'd like to clear some space on my home box to install FC4 and play around with this build to see why it fails. 4. Finally, the "Host requirements bug" 1598 is something that's always bothered me. Chris did some nice work on this, and I don't see a good reason why it shouldn't be in the book. -- Dan -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
