On Sun, Aug 11, 2013 at 05:33:53PM -0500, Bruce Dubbs wrote:
> 
> No ideas, but I decided to recheck.  I have a version of SVN I can 
> chroot into and did:
> 
> tar -xf bison-3.0.tar.xz
> cd bison-3.0
> ./configure --prefix=/usr
> echo '#define YYENABLE_NLS 1' >> lib/config.h
> make
> make check
> 
> The make completed without any warnings or errors.
> The check took a little time and skipped a few packages due to not 
> having java, doxygen, etc, but I got:
> 
> 409 tests were successful.
> 23 tests were skipped.
> 
 See my reply to myself on that - it looks as if something in my
environment is breaking it (able to build manually), but even then
the tests fall apart.

> There is really nothing special about building bison.  I was able to 
> build it without any virtual file systems mounted.  It just does 
> read-process-write.  What it does is implement the old 'yet another 
> compiler compiler' (e.g. gnu yacc) to take a specific syntax and write 
> out code to be compiled.   It's original purpose, along with lex (flex), 
> was compiler construction.
> 
> I just did:
> 
> touch lib/pipe2-safer.c
> make V=1
> 
> depbase=`echo lib/pipe2-safer.o | sed 's|[^/]*$|.deps/&|;s|\.o$||'`;\
> gcc -std=gnu99    -I. -Ilib -I. -I./lib   -g -O2 -MT lib/pipe2-safer.o 
> -MD -MP -MF $depbase.Tpo -c -o lib/pipe2-safer.o lib/pipe2-safer.c &&\
> mv -f $depbase.Tpo $depbase.Po
> 
> You should be able to do:
> 
> gcc -std=gnu99 -Ilib -c -o pipe-safer.o  lib/pipe2-safer.c
> 
> in chroot without error.
> 
>    -- Bruce
> 
 In the directory where I'd been trying the scripted build, that gcc
command works without apparent error - when run by hand.

 Thanks for the comments, I'll try with 'make V=1' to see if
anything shows up. [ /me burns some cpu cycles ... ]

 Nothing I can immediately notice from V=1 - losts of GNULIB and
other defines, but none seem to reference any variable name in my
scripts.  It fails with the same error (lib/unistd.h line 603) but
from src/AnnotationList.c.

 Attaching the log, in case anybody is motivated to look at it, but
I'm giving up for the moment.  Current summary - something in my
scripts or environment seems to be causing the breakage, but even
when I built it by hand the tests fail.

ĸen
-- 
das eine Mal als Tragödie, das andere Mal als Farce
-- 
http://linuxfromscratch.org/mailman/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/lfs/faq.html
Unsubscribe: See the above information page

Reply via email to