On 11/02/2019 22:49, Bruce Dubbs via lfs-dev wrote: > On 02/11/2019 03:19 PM, Pierre Labastie via lfs-dev wrote: >> On 11/02/2019 19:49, Bruce Dubbs via lfs-dev wrote: >>> On 02/11/2019 12:10 PM, Douglas R. Reno via lfs-dev wrote: >>> >>>> While we're here, can we please add the C.UTF-8 locale? That will resolve >>>> issues with the Gjs test suite in BLFS. >>> >>> What is the full localedef line? > >> I've not received the 2 previous messages. I've read them on the archives. >> >> For the /dev/tty issue, /dev/tty is available in chroot, but maybe it is >> "lost" when running the "su ..." command. Same thing for the "cannot set >> terminal process group" message: try running: >> su nobody -s /bin/bash -c "PATH=$PATH sh" >> in chroot, and you'll get that message. > > The message I get is 'su: must be run from a terminal', but that is as a > regular user. > > Running as root, I get: > > sh: cannot set terminal process group (-1): Inappropriate ioctl for device > sh: no job control in this shell > > But it seems to work: > > sh-5.0$ whoami > nobody > > >> But adding the locales, making the temporarylinks, > > I'm not sure which ones? chmod, touch, /usr/bin/printf, mkdir > > Anything else?
No, unless I've done something I don't remember. > > And another thought. Should we use a sed to remove the path from the places > that hardcoded path where appropriate? We could do that. I think it would amount to: sed -i '/PATH/s/usr/tools/' tests/execscript sed -i 's@/usr@/tools@' tests/intl2.sub But maybe I miss some of them... > >> and changing the book command (adding HOME): >> su nobody -s /bin/bash -c "PATH=$PATH HOME=/mnt make tests" > > What about HOME=/home ? Is anything written? Yes, much better. No nothing is written. The tests just need the ability to run "cd". So the directory must have "x" permission to "others". > >> then all but three tests pass. >> >> I'd say the command for C.UTF-8 is: >> localedef -i POSIX -f UTF-8 C.UTF-8 > > Do we care about the warnings: > > $ sudo localedef -i POSIX -f UTF-8 C.UTF-8 > [warning] No definition for LC_PAPER category found > [warning] No definition for LC_NAME category found > [warning] No definition for LC_ADDRESS category found > [warning] No definition for LC_TELEPHONE category found > [warning] No definition for LC_MEASUREMENT category found > [warning] No definition for LC_IDENTIFICATION category found > > Perhaps 'localedef -i POSIX -f UTF-8 C.UTF-8 2> /dev/null' > would be cleaner... Also: localedef -i ja_JP -f SHIFT_JIS ja_JP.SJIS generates a warning which could be sent to /dev/null. BTW, gawk wants ja_JP.UTF-8, otherwise a couple of tests are skipped (no failure) localedef -i ja_JP -f UTF-8 ja_JP.UTF-8 For other failures (not investigated yet): one failure in perl: ext/GDBM_File/t/fatal one failure in inetutils (only on i686): libls.sh 5 failures in libtool: Standalone Libtldl (123 124 125 126) and Subproject Libtldl (130) Ah, I see they are mentioned in the book. Others, maybe, I've stopped at autoconf... As you said, I'm not sure we want to fix everything... I have reported because test failures may be the symptom of something going wrong with our build. And otherwise, they should be mentioned in the book... Pierre -- http://lists.linuxfromscratch.org/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
