RE: [xml] XPath Infinities

["Ray Strode" <halfline hawaii rr com>]

> Actually, the filed bug report does not mention this.
Yeah, it does :-). It's not explicitly in the summary, 
but it is where it belongs in "OS Details" (or something like 
that, I don't remember exactly what the category is called).

>  double SetNaN()
>  {
>    return 0.0 / 0.0;
>  }
Actually, that may segfault on RedHat, too.  It's very specific,
it only allows it when the variable is being initialized to it.
It SIGFPE's any other way.  So double NaN = 0.0/0.0 will work, 
but double NaN = SetNaN(), double NaN = 0/0,and double NaN; NaN /= 0.0, 
won't.

> Ok, then my patch is appropriate for Digital Unix, and it does not disturb
> the RedHat workaround.
Okay :-).  I guess my last question is did you try it without your patch and 
it core dumped?  If so, forget I ever brought any of this up.

> The problem still existed (exits?) on Digital Unix. The patch to configure.in
> was to deal with the native C compiler. The patch to xpath.c was to deal with
> GCC on Digital Unix.
Ok, if that's the case then I withdrawl any objections I had to your patch.  
Although that means keep the #ifdef __osf__ instead of __alpha__ like we were
talking about.  Sorry to waste bandwidth over something so silly, when I shoulda
just not said anything at all :-).

--Ray