gcc43 does not compile

L

laufdi

Code: 
cc -c -DHAVE_CONFIG_H -O2 -fno-strict-aliasing -pipe -march=prescott -I/usr/local/include -I. -I../.././../gcc-4.3-20090705/libiberty/../include  -W -Wall -Wwrite-strings -Wc++-compat -Wstrict-prototypes -pedantic  ../.././../gcc-4.3-20090705/libiberty/fibheap.c -o fibheap.o
../.././../gcc-4.3-20090705/libiberty/fibheap.c: In function 'fibheap_delete_node':
../.././../gcc-4.3-20090705/libiberty/fibheap.c:258: error: 'LONG_MIN' undeclared (first use in this function)
../.././../gcc-4.3-20090705/libiberty/fibheap.c:258: error: (Each undeclared identifier is reported only once
../.././../gcc-4.3-20090705/libiberty/fibheap.c:258: error: for each function it appears in.)
gmake[3]: *** [fibheap.o] Fehler 1
gmake[3]: Leaving directory `/usr/ports/lang/gcc43/work/build/build-i386-portbld-freebsd7.2/libiberty'
gmake[2]: *** [all-build-libiberty] Fehler 2

any idea?
jdk16 doesn't compile either, probably related?
 
Clean lang/gcc43, update the ports tree and try installing again.

As for java/diablo-jdk16, I successfully installed it a short while ago. The source files need to be fetched manually before the installation begins though.

Update: If there are any compiler optimizations in /etc/make.conf, try taking them out.
 
Successfully built lang/gcc43. The port should be fine.

My environment should be different from yours though.

Code: 
$ uname -a
FreeBSD domain.tld 7.1-RELEASE-p6 FreeBSD 7.1-RELEASE-p6 #0: Wed Jun 10 21:13:14 CST 2009     root@domain.tld:/usr/obj/usr/src/sys/KERNEL  amd64

Nothing in /etc/make.conf that's related to compilation (CPUTYPE and CFLAGS not set).

I'm using amd64, so the Java part isn't built.
 
This may not help, but I've built gcc 4.3 (or whatever the current version has been) for some time. It is a required dependence of grace, a graphics program that I use (and is quite good). So I've built it and rebuilt it, and it has never failed. That is from 4.8 to 7-STABLE, i386.

I've never used anything in make.conf that affects compiler flags -- too many ways to mess up with too little effect.

Now I've not done the compilation for the last few weeks, so things may have changed. But I would be very surprised.
 
/etc/make.conf:
Code: 
CPUTYPE?=       core2
KERNCONF=       T61
WX_UNICODE=yes
WITH_MOZILLA=firefox35
WITH_GECKO=firefox35

WITH_GTK2=yes

LOCALIZED_LANG=de 
LC_ALL=de_DE.UTF-8

WITH_TTF_BYTECODE_ENABLED= yes 
WITH_GNUGCJ=yes
WITH_OPENSSL_BASE=yes
# added by use.perl 2009-07-10 21:29:09
PERL_VERSION=5.8.9

If I comment out CPUTYPE the result is the same.


emacs-22 gives error too:

Code: 
/usr/ports/editors/emacs/work/emacs-22.3/lib-src/test-distrib.c:83: error: 'EXIT_FAILURE' undeclared (first use in this function)

the only place EXIT_FAILURE ist defined is stdlib.h, but it is not included


Code: 
FreeBSD 7.2-STABLE #0: Sat Jul 11 12:38:38 CEST 2009
 
bash-4.0.24:

Code: 
cc -c -DHAVE_CONFIG_H -DSHELL  -I/usr/local/include -I. -I/usr/ports/shells/bash/work/bash-4.0 -I../.. -I../../lib  -O2 -fno-strict-aliasing -pipe -march=prescott complete.c
complete.c: In function 'rl_username_completion_function':
complete.c:1956: error: dereferencing pointer to incomplete type
complete.c:1960: error: dereferencing pointer to incomplete type
complete.c: In function 'rl_filename_completion_function':
complete.c:1979: error: expected '=', ',', ';', 'asm' or '__attribute__' before '*' token
complete.c:1979: error: 'directory' undeclared (first use in this function)
complete.c:1979: error: (Each undeclared identifier is reported only once
complete.c:1979: error: for each function it appears in.)
 ...
*** Error code 1

Stop in /usr/ports/shells/bash/work/bash-4.0/lib/readline.


something is wrong, but what?
 
freefem++-3.0.5_1 was the problem

That was the port which had installed its config.h into /usr/local/include. Unbelievable, but the port will be removed anyway.
 
You must log in or register to reply here.
Back
Top