On 8/1/06, Dan Nicholson <[EMAIL PROTECTED]> wrote:
On 7/31/06, Ken Moffat <[EMAIL PROTECTED]> wrote:
>
> I can see the symptoms in my last LFS build log for bash (in March)
> checking whether /dev/fd is available... absent
> [ in a good build, this will report 'standard' ] but I don't have
> the glibc test results, so I don't know for sure if the error
> manifests itself with glibc-2.3.6. Hopefully, somebody has test
> results ?
My message is over on clfs-dev, but I don't think this is a big issue
for the temp bash in LFS. I looked at the source and tested my temp
bash, and the process substitution works. See bash-3.1/subst.c. There
are fallbacks for this feature, and I passed c++-types a couple weeks
ago. Greg also mentioned not having any issues with glibc-2.4 and a
temp bash with the same "absent" message.
Go Moko has seemed to locate the problem here, and it is specific to
cross-compiling. Process substitution is disabled when cross-compiling
because configure can't test for the use of named pipes. So, this
shouldn't be an issue for native building in LFS. See his message
here:
http://lists.linux-phreak.biz/pipermail/clfs-dev/2006-August/000229.html
Still, we should bug upstream about the `test -r' vs. `test -e' of
/dev/fd/* since anyone building bash unprivelaged through su won't get
the HAVE_DEV_FD feature.
--
Dan
--
http://linuxfromscratch.org/mailman/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page
