In message <200208300742.g7U7gJ609111 / sharui.nakada.kanuma.tochigi.jp>
nobu.nokada / softhome.net writes:

> > ===>  Building for ruby-1.6.7.2002.07.15
> > cc -O -pipe -march=pentiumpro -fPIC -I. -I. -c process.c
> > cc -O -pipe -march=pentiumpro -fPIC -I. -I. -c ruby.c
> > cc -O -pipe -march=pentiumpro -fPIC -I. -I. -c signal.c
> > process.c: In function `proc_getpgrp':
> > process.c:801: too many arguments to function `getpgrp'
> > process.c: In function `proc_setpgrp':
> > process.c:830: too few arguments to function `setpgrp'
> > cc -O -pipe -march=pentiumpro -fPIC -I. -I. -c sprintf.c
> > *** Error code 1
> > 1 error
> > *** Error code 2

# No -j for make when make a report like this....  Yes, we can grep
# for an actual source that have problem but none prefer to so.

> It seems the bug of autoconf 2.52.

Let me explain more....

  1. first error is because config.h doesn't define GETPGRP_VOID.
  2. second since it wrongly defines SETPGRP_VOID.

2 is well known problem as nobu described, see
http://sources.redhat.com/ml/autoconf-patches/2002-02/msg00059.html

However 1 is not so, at least on my environment configure properly
detect GETPGRP_VOID is needed.  So there may be other thing.

# But I also think that's not ruby's fault but a (newer?) bug of
# autoconf.


-- 
kjana / dm4lab.to                                August 30, 2002
Haste makes waste.