On Fri, Jul 24, 1998 at 07:29:58AM -0700, Colin Wood wrote:
> Justin R. Smith wrote:
> > 
> > 
> > I have no trouble compiling it, but get
> > 
> > "Long jump botch"
> > 
> > the first time I run it.  Apparently, the
> > i386 NetBSD setup has no such problem...
> 
> What version are you using?  The last time I tried it (on 3.0?), I got the
> same thing.  Users of other m68k ports reported the same problem.  The
> thing is, I don't believe that it should even be doing a longjmp.  That
> occurs inside of an error recovery mechanism that I don't think should
> ever get called.  To me, this indicates that something really screwy is
> going on in there somewhere, but I don't know what; the code is a bit too
> dense for me to go and work my way through it at the moment :-(
> 
> If anyone has gotten it to work on an m68k-based system, please let me
> know...

I havent found time to work at it. But my guess is that 

a) postgres assumes packed structures

and I know that

b) m68k compilers align structure components at word boundaries

Try to add __attribute__((packed)) behind each structure declaration,
e.g.

struct foo {
	char bar;
	int  fum;
} __attribute__((packed));

Regards,
	-is