On Wed, Mar 10, 1999 at 02:36:46PM +0100, Reiner Dassing wrote:
> Jan Kim wrote:
[failing int regression tests]
> > This looks strange to me -- from replies, which I got from this list
> > some time ago, I know that a failing int8 test is not necessarily a
> > problem, but I'd be concerned if int2 and int4 don't work. What are
> > the differences between the actual and the expected results?
> >
>
> There are NO differences in numbers but in error explanations:Expected was "Math
> result not representable"; the actual result is "Result too large"
Ok, I've seen that too, this should not be a problem.
> The same effect can be seen wit float8. That means float8 is ok also.
>
> But, I do not understand the not usable int8 numbers. Alphas have the
> representation
> of 64 bit wide integer representation using long int as the type.
>
> Does anybody know where to search for a possible patch?
If it works ok, I don't see a need to patch anything. On an SGI Octane
64 bit system as well as Alpha), int8 for me had the "interesting"
behaviour to print out some constant, large number no matter what the
actual contents of the entity was. I asked on the list, and I got a
reply (from Bruce Momjian, IIRC) saying that a problem with int8 won't
have any effect on the other workings of PostgreSQL.
Greetinx, Jan
--
+- Jan T. Kim -------------------------------------------------------+
| email: [EMAIL PROTECTED] |
| WWW: http://www.mpiz-koeln.mpg.de/~kim/ |
*-----=< hierarchical systems are for files, not for humans >=-----*
