Bruce Dubbs wrote: > Bryan Kadzban wrote: >> Andrew Benton wrote: >>> It seems to me that we should remove udev from LFS and point >>> anyone who needs it at the systemd page in BLFS. They've shown >>> that they're integrating udev more tightly with systemd, we >>> should move away from it. >> Let's see what they say when I post the configure.ac / Makefile.am >> patch that I'm about to test out the effects of, using the .tar.xz >> archive that Bruce posted a couple days ago. >> >> Because I might just be overoptimistic, but given the success I had >> with sys/capability.h and dbus, I bet we can instead just upgrade >> to udev-185 (once it's released), and pass --disable-systemd or >> --enable-udev-only or whatever the option ends up being called. >> Looks like a couple "make install" overrides will be needed too, to >> avoid creating some directories, but that's not a huge deal >> either. >> >> >> (But I do have to test this on an actual chapter-6 setup before >> sending the patch. And the patch as-is won't work well for chapter >> 6; we'd have to change it to also modify configure and >> Makefile.in.) > > :( I don't remember that conversation but I was obviously involved. > > In BLFS we just changed the test: > > sed -i -e '21s/EXPECT_RETURN=1/EXPECT_RETURN=0/' > check/check-cmd-options
(I suspect this was meant to reply to the other thread? :-) ) If the same issue is still happening, then I think that's just covering up the bug. > and then said: > > The make phase is known to fail if the configure option > '--with-installed-popt' is used with popt-1.16. Yeah, I'd say that's an indication that the popt distributed with pkg-config is busted -- the EXPECT_RETURN sed makes the testsuite expect a failure return status, but the test program actually succeeds with a system popt that doesn't have the longArg bug. > Neither the make or the test fail for me now with > --with-installed-popt. With the BLFS sed, or without? I bet the BLFS sed makes the test appear to fail if you use --with-installed-popt. :-) And I bet that --without-installed-popt can be made to work using the info in that thread. Because it looks like Andrew Benton came up with a pkg-config sed, as well, a message later: http://www.linuxfromscratch.org/pipermail/lfs-dev/2011-September/065067.html Might be worth a shot. (If you have a system to test it on; don't go build something just for this.) OTOH that sed, or an equivalent, might have already been applied to pkg-config, or maybe they pulled in a newer version of popt.
signature.asc
Description: OpenPGP digital signature
-- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
