qnx6 patches

[David Shaw]

On Sun, Apr 15, 2007 at 05:00:26PM -0400, Sean Boudreau wrote:
> On Sun, Apr 15, 2007 at 02:58:33PM -0400, David Shaw wrote:
> > On Fri, Apr 13, 2007 at 12:02:57PM -0400, Sean Boudreau wrote:
> > > 
> > > Hello:
> > > 
> > > Please find attached the QNX6 diffs I've been using
> > > for gnupg-1.4.6.  I think most of these may be
> > > applicable to other POSIX strict environments.
> > 
> > Thank you for the diffs.  The reason for some of the changes are
> > clear, and I've started to apply them.  Some of the others I have
> > questions about:
> > 
> > > --- acinclude.m4.orig	2006-10-02 11:15:43.000000000 -0400
> > > +++ acinclude.m4	2007-03-28 09:13:07.000000000 -0400
> > > @@ -416,6 +416,42 @@
> > >                  #include <sys/types.h>
> > >                  #include <fcntl.h>
> > >  
> > > +                /* This mess was copied from the GNU getpagesize.h.  */
> > > +                #ifndef HAVE_GETPAGESIZE
> > 
> > (etc).  Can you clarify what this change is intended to address?
> > Presumably getpagesize, but can you show me what did not work in the
> > current code?
> 
> Yes, getpagesize() which qnx6 doesn't have; however we do
> have sysconf(_SC_PAGESIZE).  I was getting the
> HAVE_BROKEN_MLOCK define due to its undefined reference.
> The above is similar to what autoconf does for mmap().

Ok, I put a fix in for this (and then Werner fixed my fix!)  Anyway,
it's not the same as the patch, but the end result should be the same.
It was more than qnx: the code wouldn't work on any platform that
didn't have getpagesize().

> > I assume the various <strings.h> changes are for strcasecmp().
> 
> Yes.

Fixed as well.  The keyserver/*.c files were actually better as
ascii_strcasecmp(), so that doesn't even need strings.h.

Things should be okay for qnx6 without any special patches now.

David