On 03/06/2015 05:59 PM, Simon Sobisch wrote:
> GnuCOBOL currently only works with mpir if gmpcompat is used.
Could you perhaps elaborate your problem (?) in configuring MPIR with
'--enable-gmpcompat'?
I mean after all it's there exactly to avoid the problems you mention
below, such as having to include different headers or linking against
different(ly named) libraries.
(There are probably issues with proper library versioning on Cygwin, but
these aren't limited to GMP/MPIR I'd say.)
-leif
> Because of
> some items in our central header file libcob.h being mpz_t we had gmp.h
> included - the definitions needed for mpz_t are the only reason
> including it there.
>
> There are four possible ways I can think of to change that:
>
> 1) create libob.h from libcob.h.in during configure, placing either
> gmp.h or mpir.h in - biggest disadvantage is that you'd always need to
> create libcob.h if you use a source/svn tarball and add the creation of
> libcob.h to the vc-builds
>
> 2) use gmp.h or mpir.h depending on a define which is set during configure
> #ifdef HAVE_GMP_H
> #include <gmp.h>
> #elif defined (HAVE_MPIR_H)
> #include <mpir.h>
> /* maybe add something here like
> #else
> #pragma error "either HAVE_GMP_H or HAVE_MPIR_H have to be defined"
> */
> #endif
> Biggest disadvantage: external programs would have to add the define and
> therefore may be broken with the changed version
>
> 3) use mpir.h depending on a define which is set during configure, use
> gmp.h in all other cases
> #ifndef HAVE_MPIR_H
> #include <gmp.h>
> #else
> #include <mpir.h>
> #endif
> Biggest disadvantage: if GnuCOBOL was build with mpir while gmp.h
> (version 4.1 or version 6+) is available there may occur compatibility
> problems - at worst during runtime only (Is there any knowledge about that?)
>
> 4) use gmp.h or mpir.h depending on a define which is set during
> configure, if both defines are not set add the definition for mpz_t.
> #ifdef HAVE_GMP_H
> #include <gmp.h>
> #elif defined (HAVE_MPIR_H)
> #include <mpir.h>
> #else
> /* Definition of mpz_t (used in common.h)
> borrowed from MPIR 2.6.0
> */
> #ifndef __GNU_MP__
> typedef struct
> {
> int _mp_alloc; /* Number of *limbs* allocated and pointed
> to by the _mp_d field. */
> int _mp_size; /* abs(_mp_size) is the number of limbs the
> last field points to. If _mp_size is
> negative this is a negative number. */
> unsigned int *_mp_d; /* Pointer to the limbs. */
> } __mpz_struct;
> #endif /* __GNU_MP__ */
> typedef __mpz_struct mpz_t[1];
> #endif
> Biggest disadvantage: The definition of mpz_t may (???) not be correct
> on all systems with all versions of gmp.h / mpir.h. Same possible
> compatibility problems as with 3)
>
> I'd like to read about your thoughts and good-practices from other projects.
> Thank you for your time to answer.
>
> Simon
--
You received this message because you are subscribed to the Google Groups
"mpir-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to mpir-devel+unsubscr...@googlegroups.com.
To post to this group, send email to mpir-devel@googlegroups.com.
Visit this group at http://groups.google.com/group/mpir-devel.
For more options, visit https://groups.google.com/d/optout.
