🚀 go-pugleaf

In article <2198vm$2m6@dux.dundee.ac.uk>, ndevnani@dee.mcs.dundee.ac.uk (ndevnani@uk.ac.dundee.mic) writes:
> Hi All,
>       We're using gcc with the -ansi switch  on SunOS 4.1.2 and getting
>       compiler errors in the math.h header file which is included in the
>       source.
>       Does anyone have any idea what this could be? My initial reaction was
>       math.h with cc (which is not ansi) is being included instead of
>       header files supplied with gcc and causing problems but the more
>       I think about it the more I'm unconvinced as there are no problems
>       with the other commmon header files.  
>       If anyone has any ideas, I'd be grateful.
>   
>       Many thanks,
>       Nigel Devnani, University of Dundee, Scotland, UK.
>       e:mail ndevnani@uk.ac.dundee.mi

We had the same problem here some time ago, and i found out the following facts about it:

In math.h (or some file included in there, i am not so sure about this) there
are several (co-)processor specific definitions. These are switched by "#ifdef"s 
for "sparc" "68k" or so. Those Definitions are removed from the list of predefinitions
of the gnu cpp if "-ansi" is in effect, along with any predefinitions not starting
with "__". You may just leave "-ansi" away if you do not really need it or add the 
definitions (-Dsparc) in the compiler call (CFLAGS= ... -Dsparc).

The option "-ansi" is only necessary if you want to omit all ansi supersets for 
compatibility or need e.g. trigraphs (do you?). I could compile all ansi-compliant 
programs without this option up to now. Leaving this option away does not mean that 
gcc behaves like a K&R Compiler!

Wolfram Wagner                          Email: ww@mpi-sb.mpg.de           
Max-Planck-Institut fuer Informatik     Tel.:  +49-681/302-5403
Im Stadtwald                            Fax:   +49-681/302-5401
66123 Saarbruecken GERMANY