Mail Archives: djgpp/2004/09/18/15:51:48
This is still about binaries and making them, but I have finished
with the binutl-2.15 package. This is really about how well
djgpp-2.04_alpha, and recent GNU packages including gcc-3.4.2
behave on a computer running Windows XP Home Edition. If I were
wiser I should have chosen a better thread line. If I wanted
just to compile and use the packages, I would choose a different
operating system. This system, off the shelf, with XP and DJGPP
gets good marks. But there are problems. That's why were're here.
diffutil.281 had that interesting problem with the two dead
symlinks. "ls -l" showed the link attribute of the dead file
but nothing else. After applying "dtou," it became transparent
to "ls -l!" Surely you will not begrudge me the "Ah! Ha!" over
this serendipitous fix.
Taking up next the package "gmp-4.12," my last post reported a
score of dead links compiling as it was prepared, except "configure"
was left free to choose the processor type. It found a pentium4.
The actual count of linked files was 20h. };=) It also found five
errors running "make -k" which are reported below.
There was that persistant warning about "-mpcu" which had to be
changed to "-mtune" in all the Makefiles in all the subdirectories.
It only needs to be changed once in "configure.in" if the package
can be reconfigured. I ran "make distclean" which was probably
a mistake. This was before I learned about "autoreconfig -f".
Anyway it didn't work. It reported required macros which were
undefined, and two instances of subdirectory "Makefiles.am" which
had CFLAGS instead of AM_CFLAGS. I hope now that this was not
intentional since I replaced CFLAGS with AM_CFLAGS. In any event
"doc/configuration" contains a number of things to be done which
I did, including getting recent files "config.guess" and
"config.sub" from GNU. In retrospect, I should have searched for
new "configfsf.guess" and "configfsf.sub" files as well.
"autoreconfig -f" now ran successfully. There were -mtune's in
all the Makefile's. What seems to have been lost in the configure
process is capability to choose the archecture of the processor.
Given the freedom to select, it settled on "-mtune=i486 -march=i486"
There were now 55 linked files. "dtou @linkfile.rsp" did them all,
whether they needed it or not. The first one reached in the
compilation certainly did. Before dtou was applied the symlinked
files now got a normal unix format response from "ls -l" (with -->
pointing to the primary file). If I can get past the errors the
second time around should go better.
Below are the errors and excerpts from the files. This is
gcc-3.4.2! These errors were generated from the original package
and from the reconfigured one as well. What is the fix??
Original package, running configure with HOST=i386-pc-msdosdjgpp.
"-mcpu" was replaced by "-mtune" in the Makefile's.
bash-2.05b$ redir -eo -x -t make -k > error.cache
------------------------------------------------------------------
gcc -DHAVE_CONFIG_H -I. -I. -I. -D__GMP_WITHIN_GMP -g -O2 \
-fomit-frame-pointer -mtune=i486 -march=i386 -c mpz/aorsmul.c
mpz/aorsmul.c:44: error: conflicting types for '__gmpz_aorsmul'
mpz/aorsmul.c:39: error: previous declaration of '__gmpz_aorsmul' \
was here
mpz/aorsmul.c:44: error: conflicting types for '__gmpz_aorsmul'
mpz/aorsmul.c:39: error: previous declaration of '__gmpz_aorsmul' \
was here
make.exe[2]: *** [mpz/aorsmul.lo] Error 1
from mpz/aorsmul.c
39: static void __gmpz_aorsmul _PROTO ((REGPARM_3_1 \
(mpz_ptr w, mpz_srcptr x, mpz_srcptr y, mp_size_t sub))) \
REGPARM_ATTR (1);
40: #define mpz_aorsmul(w,x,y,sub) __gmpz_aorsmul (REGPARM_3_1 \
(w, x, y, sub))
41:
41: static void
4 mpz_aorsmul (mpz_ptr w, mpz_srcptr x, mpz_srcptr y, mp_size_t sub)
44: {
-----
gcc -DHAVE_CONFIG_H -I. -I. -I. -D__GMP_WITHIN_GMP -g -O2 \
-fomit-frame-pointer -mtune=i486 -march=i386 -c mpz/aorsmul_i.c
mpz/aorsmul_i.c:62: error: conflicting types for '__gmpz_aorsmul_1'
./gmp-impl.h:628: error: previous declaration of '__gmpz_aorsmul_1' \
was here
mpz/aorsmul_i.c:62: error: conflicting types for '__gmpz_aorsmul_1'
./gmp-impl.h:628: error: previous declaration of '__gmpz_aorsmul_1' \
was here
make.exe[2]: *** [mpz/aorsmul_i.lo] Error 1
from gmp-impl.h
628: void __gmpz_aorsmul_1 _PROTO ((REGPARM_3_1 (mpz_ptr w, \
mpz_srcptr u, mp_limb_t v, mp_size_t sub))) REGPARM_ATTR(1);
from mpz/aorsmul_i.c
61: mpz_aorsmul_1 (mpz_ptr w, mpz_srcptr x, mp_limb_t y, \
mp_size_t sub)
62: {
-----
gcc -DHAVE_CONFIG_H -I. -I. -I. -D__GMP_WITHIN_GMP -g -O2 \
-fomit-frame-pointer -mtune=i486 -march=i386 -c mpz/cfdiv_q_2exp.c
mpz/cfdiv_q_2exp.c:34: error: conflicting types for \
'__gmpz_cfdiv_q_2exp'
mpz/cfdiv_q_2exp.c:29: error: previous declaration of \
'__gmpz_cfdiv_q_2exp' was here
mpz/cfdiv_q_2exp.c:34: error: conflicting types for \
'__gmpz_cfdiv_q_2exp'
mpz/cfdiv_q_2exp.c:29: error: previous declaration of \
'__gmpz_cfdiv_q_2exp' was here
make.exe[2]: *** [mpz/cfdiv_q_2exp.lo] Error 1
from mpz/cfdiv_q_2exp.c
29: static void __gmpz_cfdiv_q_2exp _PROTO ((REGPARM_3_1 \
(mpz_ptr w, mpz_srcptr u, unsigned long cnt, int dir))) \
REGPARM_ATTR (1);
30: #define cfdiv_q_2exp(w,u,cnt,dir) __gmpz_cfdiv_q_2exp \
(REGPARM_3_1 (w,u,cnt,dir))
31:
32: static void
33: cfdiv_q_2exp (mpz_ptr w, mpz_srcptr u, unsigned long cnt, \
int dir)
34: {
-----
gcc -DHAVE_CONFIG_H -I. -I. -I. -D__GMP_WITHIN_GMP -g -O2 \
-fomit-frame-pointer -mtune=i486 -march=i386 -c mpz/cfdiv_r_2exp.c
mpz/cfdiv_r_2exp.c:37: error: conflicting types for \
'__gmpz_cfdiv_r_2exp'
mpz/cfdiv_r_2exp.c:32: error: previous declaration of \
'__gmpz_cfdiv_r_2exp' was here
mpz/cfdiv_r_2exp.c:37: error: conflicting types for \
'__gmpz_cfdiv_r_2exp'
mpz/cfdiv_r_2exp.c:32: error: previous declaration of \
'__gmpz_cfdiv_r_2exp' was here
make.exe[2]: *** [mpz/cfdiv_r_2exp.lo] Error 1
from mpz/cfdif_r_2exp.c
32: static void __gmpz_cfdiv_r_2exp _PROTO ((REGPARM_3_1 \
(mpz_ptr w, mpz_srcptr u, unsigned long cnt, int dir))) \
REGPARM_ATTR (1);
33: #define cfdiv_r_2exp(w,u,cnt,dir) __gmpz_cfdiv_r_2exp \
(REGPARM_3_1 (w, u, cnt, dir))
34:
35: static void
36: cfdiv_r_2exp (mpz_ptr w, mpz_srcptr u, unsigned long cnt, \
int dir)
37: {
-----
gcc -DHAVE_CONFIG_H -I. -I. -I. -D__GMP_WITHIN_GMP -g -O2 \
-fomit-frame-pointer -mtune=i486 -march=i386 -c mpq/aors.c
mpq/aors.c:33: error: conflicting types for '__gmpq_aors'
mpq/aors.c:27: error: previous declaration of '__gmpq_aors' \
was here
mpq/aors.c:33: error: conflicting types for '__gmpq_aors'
mpq/aors.c:27: error: previous declaration of '__gmpq_aors' \
was here
make.exe[2]: *** [mpq/aors.lo] Error 1
from mpq/aors.c
27: static void __gmpq_aors _PROTO ((REGPARM_3_1 (mpq_ptr w, \
mpq_srcptr x, mpq_srcptr y, void (*fun) _PROTO ((mpz_ptr, \
mpz_srcptr, mpz_srcptr))))) REGPARM_ATTR (1);
28: #define mpq_aors(w,x,y,fun) __gmpq_aors \
(REGPARM_3_1 (w, x, y, fun))
29:
30: static void
31: mpq_aors (mpq_ptr rop, mpq_srcptr op1, mpq_srcptr op2,
32: void (*fun) _PROTO ((mpz_ptr, mpz_srcptr, mpz_srcptr)))
33: {
-----
gcc -DHAVE_CONFIG_H -I. -I. -I. -D__GMP_WITHIN_GMP -g -O2 \
-fomit-frame-pointer -mtune=i486 -march=i386 -c mpf/ceilfloor.c
mpf/ceilfloor.c:36: error: conflicting types for \
'__gmpf_ceil_or_floor'
mpf/ceilfloor.c:31: error: previous declaration of \
'__gmpf_ceil_or_floor' was here
mpf/ceilfloor.c:36: error: conflicting types for \
'__gmpf_ceil_or_floor'
mpf/ceilfloor.c:31: error: previous declaration of \
'__gmpf_ceil_or_floor' was here
make.exe[2]: *** [mpf/ceilfloor.lo] Error 1
from mpf/ceilfloor.c
31: static void __gmpf_ceil_or_floor _PROTO ((REGPARM_2_1 \
(mpf_ptr r, mpf_srcptr u, int dir))) REGPARM_ATTR (1);
32: #define mpf_ceil_or_floor(r,u,dir) __gmpf_ceil_or_floor \
(REGPARM_2_1 (r, u, dir))
33:
34: static void
35: mpf_ceil_or_floor (mpf_ptr r, mpf_srcptr u, int dir)
36: {
-----
Frank
-----------------------------------------------------------------------
This mail was sent through Wilkes Webmail: http://webmail.wilkes.edu
Wilkes Webmail is using IMP: http://horde.org/imp/
- Raw text -