operator should be followed by a macro argument name

operator should be followed by a macro argument name

[John Ling]

I have the following setup:

$ uname -a
AIX server1 2 5 000B079F4C00
$ gcc -v
Reading specs from /usr/local/lib/gcc-lib/powerpc-ibm-aix4.3.2.0/2.95.3/specs
gcc version 2.95.3 20010315 (release)

I am trying to build GCC version 3.4.4 using the following configuration:

./configure
make

and I get the following errors.  I appreciate any suggestions anyone might
have.

make[1]: Entering directory
`/flash/local/software/gcc/downloads/gcc-3.4.4/libiberty'
if [ x"" != x ]; then \
  gcc -c -DHAVE_CONFIG_H -g -O2 -I. -I./../include  -W -Wall -Wtraditional
-pedantic  floatformat.c -o pic/floatformat.o; \
else true; fi
gcc -c -DHAVE_CONFIG_H -g -O2 -I. -I./../include  -W -Wall -Wtraditional
-pedantic floatformat.c -o floatformat.o
In file included from floatformat.c:27:
/usr/local/lib/gcc-lib/powerpc-ibm-aix4.3.2.0/2.95.3/include/math.h:1070:
`#' operator should be followed by a macro argument name
/usr/local/lib/gcc-lib/powerpc-ibm-aix4.3.2.0/2.95.3/include/math.h:1062:
unterminated `#if' conditional
make[1]: *** [floatformat.o] Error 1
make[1]: Leaving directory
`/flash/local/software/gcc/downloads/gcc-3.4.4/libiberty'
make: *** [all-libiberty] Error 2

Thanks,
John Ling

Re: operator should be followed by a macro argument name

[Ian Lance Taylor]

"John Ling" <jling@bioinformatics.ubc.ca> writes:

> /usr/local/lib/gcc-lib/powerpc-ibm-aix4.3.2.0/2.95.3/include/math.h:1070:
> `#' operator should be followed by a macro argument name

What is line 1070 in the file 
    /usr/local/lib/gcc-lib/powerpc-ibm-aix4.3.2.0/2.95.3/include/math.h
?

This may be a bug in the way gcc 2.95.3 fixed the header file.  If so,
your best bet is probably going to be to patch it by hand.

Ian

Re: operator should be followed by a macro argument name

[John Ling]

Line 1070 is as follows:

1070:  #ifndef __cplusplus
1071:          ((_class(__x) == FP_SNAN) || (_class(__x) == FP_QNAN) ? \
1072:  #endif

It looks fine to me.  But i could try removing these ifndef/endif lines.

Is gcc-lib a separate package?  I notice it refers to aix4.3.2.0 yet my 
aix is 5.2.0.0.  Perhaps that could be a reason why its not working.  If 
its a separate package then maybe i should install that first.

John

Ian Lance Taylor wrote:
> "John Ling" <jling@bioinformatics.ubc.ca> writes:
> 
> 
>>/usr/local/lib/gcc-lib/powerpc-ibm-aix4.3.2.0/2.95.3/include/math.h:1070:
>>`#' operator should be followed by a macro argument name
> 
> 
> What is line 1070 in the file 
>     /usr/local/lib/gcc-lib/powerpc-ibm-aix4.3.2.0/2.95.3/include/math.h
> ?
> 
> This may be a bug in the way gcc 2.95.3 fixed the header file.  If so,
> your best bet is probably going to be to patch it by hand.
> 
> Ian

Re: operator should be followed by a macro argument name

[Ian Lance Taylor]

John Ling <jling@bioinformatics.ubc.ca> writes:

> Line 1070 is as follows:
> 
> 1070:  #ifndef __cplusplus
> 1071:          ((_class(__x) == FP_SNAN) || (_class(__x) == FP_QNAN) ? \
> 1072:  #endif
> 
> It looks fine to me.  But i could try removing these ifndef/endif lines.
> 
> Is gcc-lib a separate package?  I notice it refers to aix4.3.2.0 yet
> my aix is 5.2.0.0.  Perhaps that could be a reason why its not
> working.  If its a separate package then maybe i should install that
> first.

gcc-lib is not a separate package.  The files in
gcc-libl/TARGET/VERSION/include are created by fixing the header files
on the system when gcc is installed.  This process needs to happen
because many systems have header files which do not by themselves work
with gcc.

It may be a problem that you have a fixed version of the AIX 4.3.2.0
header files when you are using AIX 5.2.0.0.  The simplest way to fix
that is to reinstall gcc, assuming you still have the source code.  Or
to find a binary package for AIX 5.2.0.0.

Those lines in the header file look fine.  Since the error was

> >>/usr/local/lib/gcc-lib/powerpc-ibm-aix4.3.2.0/2.95.3/include/math.h:1070:
> >>`#' operator should be followed by a macro argument name

I see now that gcc thinks that line 1070 is in a macro definition.  Is
there perhaps a backslash at the end of line 1069, where line 1069 is
in a #define?

Ian