CLN + GiNaC on MinGW

[Richard B. Kreckel]

Hi!

On Fri, 31 Oct 2003, Jonathan Brandmeyer wrote:
> I am attempting to build CLN (and later, GiNaC) on my Windows XP machine
> with Mingw.  I am using g++ 3.2.3 and the latest binary utilities.
>
> I have a build of GMP 4.2.1 for use as a backend for CLN (installed
> under c:\msys\1.0\local\).
>
> CLN was configured with CPPFLAGS=-I/usr/local/include
> LIBS=/usr/local/lib/libgmp-3.dll CXXFLAGS="-O2 -finline-functions
> -fomit-frame-pointer -march=i686 -finline-limit=1000" ./configure
> --with-gmp.
>
> The build progressed and after a couple hours the link failed for the
> shared library.  It looks a lot like what was described on ginac-devel
> when a fellow had problems on Alpha (thread begins here:
> http://www.ginac.de/lists/ginac-devel/msg00091.html), except that I am
> using a relatively modern toolchain.
>
> The linker errors are of the form:
> cl_t_c2.lo(.text+0x4e):cl_t_c2.cc: undefined reference to
> `_GLOBAL__I_cl_module__cl_prin_globals__firstglobalfun'
>
> Any thoughts?
> -Jonathan Brandmeyer
>
> P.S.  I am using msys to drive the build, but I do not want to use
> Cygwin.

Sorry, no clue.  I don't know of anyone using your toolchain.  But I am
sure that it should work at least on Cygwin since we've received some
smallish patches (which've all made it into our tree).

Wait!  -finline-functions is something which is switched on with -O3
normally.  Now, I do know that there used to be such unresolved symbols
with -O3.  Since not much else is switched on with -O3 (only register
renaming), maybe that would be the culprit?  Could you please try without
-finline-functions and see how it goes?!

BTW, as of GCC-3.4 (current CVS snapshot) CLN should now compile fine with
full -O3 optimizations (which *does* make a difference, but only due to
the new -funit-at-a-time option, not due to the passes already present
in GCC-3.3).

Luck
  -richy.
-- 
Richard B. Kreckel
<Richard.Kreckel at GiNaC.DE>
<http://www.ginac.de/~kreckel/>