On Fri, Dec 04, 2020 at 03:25:20AM +0000, Ken Moffat wrote:
> On Thu, Dec 03, 2020 at 10:50:50PM +0100, Pierre Labastie wrote:
> >
>
> Now to work out what to build ;-)
>
An interim report, mainly to highlight some "That hurts!" "Well,
don't do that, then." items.
1. Putting XORG variables in envars.conf was a bad idea. First,
xorg-env still has to be edited. Second I used $XORG_PREFIX in
envars.conf instead of \$XORG_PREFIX, maybe that was releated.
Anyway, util-macros read the prefix as ORG_PREFIX (without the /$X)
and therefore installed into the current (build) directory. The
protocols installed correctly, but libXau fell over badly with the
missing macros.
2. /dev/shm and specifically the book's instructions for chroot.
When I built Xorg (including JS78) I had used my known-good mounts
for building BLFS (on sysv) :
bind /dev
mount /dev/pts
mount proc
mount sys
mount /dev/shm using a tmpfs
bind /run
That last item seems odd, but the whole process worked (odd because
mountvirtfs creates the antique /dev/shm -> /run/shm symlink which
implies /dev/shm is actually mounted on /run/shm but that /run has
overlaid that).
For proceding to firefox I dropped the shm and run parts, to include
in my firefox (etc) scripts the book's instruction to mount /shm.
Eventually I got to firefox, and no matter what I tried, the
python configure gave me the old:
raise ImportError("This platform lacks a functioning sem_open"
At that point I went back to my notes, and the -dev archive from
March and tried everything that was suggested, and a few attempts of
my own to mount/umount /dev/shm (DO NOT try that at home if you have
firefox or falkon open - need to reboot to get them back!).
Got nowhere. I had noticed that python3 got recompiled shortly
before the Makefile got to firefox. Whenever I've rebuilt it in
chroot on my own systems I've always had the LFS shm mounts.
So, to be sure I booted the jhalfs system and tried to build
firefox. It failed with the same error.
So I've now gone back to chroot with my normal mounts, rebuilt
python3 and now firefox has completed.
This means that I have not managed to verify whether or not the
book's comments on shm in mozilla packages are correct, but also
that python3 needs working shm when it is (re) built.
Same for python2, otherwise the seamonkey configure fails in the
same way.
I always used to build both pythons very early in my BLFS builds, in
chroot before the first boot, so that is not necessarily a change.
ĸen
--
To say that it (his hair) was black and bound up in a ponytail is to
miss the opportunity of using the term 'elephantine'. It was hair
with personality. -- The Thief Of Time (about the monk, Sato).
--
http://lists.linuxfromscratch.org/listinfo/alfs-discuss
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page
