Re: beta-8

[Norbert Nemec]

On Tue, May 01, 2001 at 12:54:17PM +1200, Keith Hopper wrote:
> Greetings,
>      I have put beta-8 on our sather web site
> (http://www.cs.waikato.ac.nz/sather) for anyone who wishes to use it.  All
> works well under linux, but there are a couple of things which are odd -
> not about the code- under Windows2000:-
> 
>      (1)  I can't seem to get the command ./configure cygwin to work.  The
>      shell doesn't seem to like the configure script - so I have just had
>      to manually enter the last (copy) line by hand replacing the arg by
>      'cygwin'.

configure is a make-script. The way it is called on UN*X can, of course not 
work 
for Windows - I did not think of the Windows-port at all, when I constructed 
that mechanism...

Just call 'make -f configure' instead (That's exactly what that first comment 
line implies for UN*X systems) On the long run, the whole thing should probably 
replaces by a shell script which is more portable.

>      (2)  The makefile in the test directory will not compile programs - I
>      -think- this is something to do with there being an undefined define
>      for WIN32 - only applicable under Windows, of course.  If Nobbi (who
>      wrote it) would like to try to fix it before moving onto the Gnu site
>      I would be very grateful - TIA.

The sources depend on the symbol __CYGWIN32__ being automatically defined on all
CYGWIN systems. Maybe that assumption is not correct. Just check it yourself. 
On my system, I see:

-------------
address@hidden:~/cvs/sather/system $ touch /tmp/x.c
address@hidden:~/cvs/sather/system $ gcc -E -dM /tmp/x.c
#define __linux__ 1
#define linux 1
#define __i386__ 1
#define __i386 1
#define __GNUC_MINOR__ 95
#define i386 1
#define __unix 1
#define __unix__ 1
#define __GNUC__ 2
#define __linux 1
#define __ELF__ 1
#define unix 1
address@hidden:~/cvs/sather/system $
---------------

I hope the commands line options work for you as well, so you can check whether 
__CYGWIN32__ is actually defined by the system or what symbol might better be 
used instead. Only other possiblity I see is that the preprocessor syntax 

#if !defined(__NeXT__) && !defined(__EMX__) && !defined(__NetBSD__) && 
!defined(__FreeBSD__) && !defined(__CYGWIN32__)
#  include <values.h>
#endif

(./system/c_header.h, line 43) is either incompatible or incorrect. It should 
not depend on the makefiles in any way, though.

Ciao,
Nobbi

-- 
-- ______________________________________________________
-- JESUS CHRIST IS LORD!
--          To Him, even that machine here has to obey...
--
-- _________________________________Norbert "Nobbi" Nemec
-- Hindenburgstr. 44  ...  D-91054 Erlangen  ...  Germany
-- eMail: <address@hidden>   Tel: +49-(0)-9131-204180