RE: [xml] XPath Infinities

Ray Strode wrote:

The bug report only mentioned Alpha, and I assumed that we were talking
about Digital Unix (or True64 or whatever they call it these days), so
that's what I tested.
No, it says RedHat 7.0, but good to test on that too, although Daniel 
said that it isn't as much of a priority as RedHat.

And no, __osf__ is not defined for RedHat. The native Unix on Alpha has
changed name a couple of times, but originally it was called OSF/1, and
the __osf__ define has stuck ever since.
I figured so much.

Are you using Linux on Alpha? 
Yes.  RedHat 7.0.

If so, you could try with the __alpha__
(or something similar) define instead of __osf__. You can get a list of
built-in defines for GCC with
 gcc -E -dM -x c /dev/null
Thanks; that's a cool little tip.

If you use GCC on Alpha, then 0/0 will normally result in a core dump.
Yup, but 0/0.0 doesn't (at least on RedHat 7.0).

So, you have to ignore SIGFPE on Alpha using GCC.
I don't know about Tru64, but on RedHat you don't.

Right. Instead we get a NaN.
Exactly. It will be NaN and thus the signal trapping is not required.

With C99 compilers there is a NAN macro which can be used instead.
Well, maybe that should be done, too, then.

Or are you saying that the bug still existed on OSF?
With the native C compiler, yes.
But the other part of your patch was for the native C compiler, yea? 
(I mean the -ieee stuff).  Or is xpath.diff necessary for the native C
compiler, too?


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]