libffi test

[Bryce McKinlay]

Andreas Tobler wrote:

>With a binary and with sun java I get this (as expected):
>
>[titanium:~] andreast% ./my_sqrt
>sqrt_int: 0
>sqrt_double: NaN
>sqrt_long: 0
>
>Doing it with gij I get the following:
>
>[titanium:~] andreast% gij my_sqrt
>sqrt_int: -2147483648
>sqrt_double: NaN
>sqrt_long: -9223372034707292160
>
>Is this obviously a problem in the libffi ? Or somewhere else?
>
>One strange thing I see is, the integer is the biggest negativ one
>defined. The long on the otherside is not the biggest negativ one. (-9223372036854775808)
>---
>So is this to worry about? 
>

Oh, I forgot about this. This is (I think) an endian bug in libjava 
and/or gcj -C, and occurs on Linux PPC as well. I'm not sure exactly 
where the problem is here but I'm pretty sure it isn't libffi.

regards

Bryce.