"Andrey G. Grozin" <A.G.Grozin at inp.nsk.su> writes:
> This is the first idea which comes to one's mind. But:
>
> 1. If I ./configure maxima for just one lisp (sbcl or gcl or any other),
> make succeeds. Maybe, earlier actions related to other lisps change
> something in the tree, and these changes influence behaviour of make when
> working with gcl and sbcl?
>
> 2. In 5.27.0 and many earlier version this problem did not exist. This,
> probably, means that some change in the build system between 5.27.0 and
> 5.28.0 is the cause of the problem.
>
> 3. I run make -j2, so, make may run 2 processes simultaneously. Can it be
> some kind of race condition? Looking at the log file where maxima was
> ./configured for all 6 lisps, I see lines with "binary-gcl" and
> "binary-sbcl" intermixed. So, it seems that make decided to run sbcl and
> gcl compilations in parallel. Can this lead to troubles?
FWIW, I built 5.28.0 on a 32 & a 64-bit debian testing system, with
clisp, ecl, gcl, and sbcl enabled and the builds were flawless. So I
doubt what you are seeing is down to a change in Maxima's build
scripts. The commands I ran were
./bootstrap
./configure --enable-sbcl --enable-ecl --enable-gcl --enable-clisp \
--with-default-lisp=sbcl EMACS=t --no-create --no-recursion
make clean
rm -fr src/binary-*
make && make check
I remove all the src/binary-* directories because the lisps have
generally changed since the last build...
What happens if you run make without the -j2 flag?
--
Leo Butler <l_butler at users.sourceforge.net>
SDF Public Access UNIX System - http://sdf.lonestar.org