I am using the Absoft F77 Compiler for Linux since it can handle Fortran
structures and I've run into a problem using AC_F77_LIBRARY_LDFLAGS in the
CVS autoconf as of June 6th. I had used an earlier CVS version from around
last Decemeber which worked fine, but I don't have that copy anymore... so
> | configure:1039: /usr/absoft/bin/f77 -o conftest conftest.f >&5
> | FORTRAN 77 Compiler 4.5, Copyright (c) 1987-1999, Absoft Corp.
>
> Hm, are you telling us that your compiler is always outputting
> something when it compiles?
Yes. I get tons of them during a make command.
$ /usr/absoft
$ more config.log
This file contains any messages produced by compilers while
running configure, to aid debugging if configure makes a mistake.
It was created by configure 2.14a, executed with
> ../configure
configure:595: checking for gcc
configure:617: result: gcc
configure:709: checking wh
Okay, maybe I don't know how to use CVS, but I just did this:
$ cvs -z9 -d :pserver:[EMAIL PROTECTED]:/cvs update autoconf
cvs server: Updating autoconf
P autoconf/ChangeLog
P autoconf/NEWS
P autoconf/acgeneral.m4
P autoconf/aclang.m4
P autoconf/acspecific.m4
cvs s
I am searching for dlopen in both the c and dl libraries and want
to create the variable LIBDL as either blank or -ldir. I do this
rather than add it to LIBS because my SGI linker gives me warnings
that the 'dir' library does not resolve any symbols for those
programs that do not use 'dlopen'. I
Is there a way to cleanly prevent these macros from setting
[C,F]FLAGS to "-g -O2" by default?
I am using automake/autoconf and am in a situation in which I need
to use a program to generate a header file using a template file
before I can compile the program. Has anyone encountered a similiar
situation? What is the best way to accomplish this? Suggestions.
Thanks, Paul
I just checked out the autoconf and tried to run ifnames on a short
hello world program. I am getting this error:
ifnames test.c
gawk: cmd. line:57: warning: END blocks must have an action part
My gawk is 3.03 (for both an SGI and Linux system -- same error).
Moving the { from the line bel
Based on the documentation, ifnames will output only the identifiers used
in #if directives. The CVS version dumps all the lines that start with
# rather than just the identifiers.
$ cat test.c
#include
#if HAVE_SOMETHING
#endif
int
main(int argc, char *argv[] )
{
printf("Hello, World!\n")
Using CVS autoscan to produce a configure.scan file, I noticed it used
the older 2.14 way of doing things: AC_INIT(unique-file-in-source-dir).
With autoconf 2.49b, the AC_INIT macro is documented as requiring the
PACKAGE and VERSION (from automake, I guess) and the newer macro
AC_CONFIG_SRCDIR(un
With this configure.in, I get the same warning twice.
$ cat configure.in
AC_INIT(aps,2.4,[EMAIL PROTECTED])
AC_PROG_CPP
AC_PROG_CC(cc gcc)
AC_CHECK_FUNCS(sincos)
AC_OUTPUT
$ autoconf
configure.in:3: warning: AC_PROG_CPP was called before AC_PROG_CC
configure.in:4: warning: AC_PROG_CPP was
to revert back to a specific version for each
file? I am not very familiar with cvs.
Thanks,
Paul
--
+-+-+-+-+-+-+-+-+-+-+-----+-----+
Paul Martinolich/Software Engr. e-mail:
[EMAIL PROTECTED]
Neptune Sciences, Inc.voice: (228) 688-5280
is output does not look like what is supposed to be
placed in the configure file from _AC_LANG_COMPILER_GNU. I am checking
for
a C compiler (which also calls _AC_LANG_COMPILER_GNU) before the
AC_PROG_F77
call.
I just tested the order of AC_PROG_CC and AC_PROG_F77 in a configure.in
file and it make
t;/dev/null 2>&1;
then
ac_cv_prog_g77=yes
else
ac_cv_prog_g77=no
fi])])
--
+-+-+-+-+-+-+-+-+-----+-+-+-+
Paul Martinolich/Software Engr. e-mail:
[EMAIL PROTECTED]
Neptune Sciences, Inc.voice: (228) 688-5280
NRL Remote Sensing/Ocean
; because it can't determine if it is a GNU compiler using strict
Fortran source.
To me, the best solution would be to replace the generic
_AC_COMPILER_GNU
macro from AC_PROG_F77 with the 2.14 AC_PROG_F77_GNU macro. The only
problem
with that macro maybe how different Fortran compilers handle the .fp
The autoscan from autoconf-2.49c has ugly output for the
missing macro warning.
$ ../../devel/bin/autoscan.orig
warning: missing AC_TYPE_PID_T wanted by: libsdptk/src/PGS_IO_Gen_Temp_Reference.c:675
libsdptk/src/PGS_SMF_Comp.c:5336 libsdptk/src/PGS_SMF_Comm.c:139
libsdptk/smfcompile/PGS_SMF_Co
I am wondering if the new preferred way of an empty AC_OUTPUT
and use of AC_CONFIG_FILES in autoconf-2.49c is presently
known to automake-1.4d or does automake still require the
old way of doing AC_OUTPUT. Thanks.
Paul
Unless, I am reading the manual (2.49c) wrong the following is
prefered:
AC_CONFIG_AUX_DIR([config])
to
AC_CONFIG_AUX_DIR(config)
With automake-1.4d, in the first case the configure directory is
`[config]' rather than `config'.
$ ../../devel/bin/automake
automake: configure.ac: required file
I have directory with:
lib_LIBRARIES = libjunk.a
libjunk_a_SOURCES = src/a/a.f \
src/b/b.c
run configure and gmake and get this:
$ gmake
source='src/b/b.c' object='b.o' libtool=no \
depfile='.deps/b.Po' tmpdepfile='.deps/b.TPo' \
depmode=gcc /bin/sh ../test/depcomp \
gcc -D
I am using the Absoft Fortran 6.0 compiler for Linux and the
mangling as reported by AC_F77_WRAPPERS is lower case, but
documentation indicates uppercase. And in fact, when I manually
change the config.h file to use uppper case it links correctly.
Has anyone else used this compiler and/or have a
I have a configure.ac in which I conditionally configure additional
sources that have their own ./configure scripts. Using AC_CONFIG_SUBDIRS
works great. But, now I'd like to pass some configure options that I
don't use in my top-level configure. How can I pass these to the individual
configure
Sorry, I was unclean. I'll give it another shot. I have this structure
../src
../pkgA
../pkgB
src contains my application. Platform A does not provide the third party
pkgA which I need in the application, whereas Platform B does. Ditto for
pkgB. Some platforms provide it and others do
22 matches
Mail list logo
