integer test hangs on leopard (darwin9) intel core 2 duo

the testsuite hangs in intmath.st. when canceling this the rest of the  
test suite passes.

output of running the file manually:

pingi@Macbook ~/s/s/tests> gst  intmath.st
true
true
'Shift -1 left then right and test for -1'
'Shift 1 left then right and test for 1'
'And a single bit with -1 and test for same value'
'Verify that (n bitAnd: n negated) = n'
'Verify that (n + n complemented) = -1'
'Verify that n negated = (n complemented +1)'
'LargeInteger bit logic tests passed'
64212742967590157
1216451004088320000
"Global garbage collection... done"
^Cintmath.st:206: Interrupt
(ip 86)LargePositiveInteger>>#bytesRightShift:big:
(ip 220)LargePositiveInteger>>#gcd:
(ip 8)LargeNegativeInteger>>#gcd:
(ip 98)Fraction>>#-
(ip 30)[] in Stream>>#accelerate
(ip 26)Generator>>#forkOn:
pingi@Macbook ~/s/s/tests>

pingi@Macbook ~/s/s/tests> gst -D intmath.st : log appended

pingi@Macbook ~/s/s/tests> gst -E intmath.st : log appended




_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk

no, I did not forget to attach. i was only too stupid to hit the  
button correctly. here comes the





_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk

Ingolf Jandt <[hidden email]> writes:
> the testsuite hangs in intmath.st. when canceling this the rest of the
> test suite passes.

Please install GMP and try it then.
See http://lists.gnu.org/archive/html/help-smalltalk/2008-02/msg00022.html

--
But you know how reluctant paranormal phenomena are to reveal
themselves when skeptics are present. --Robert Sheaffer, SkI 9/2003


_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk

Stephen Compall wrote:
> Ingolf Jandt <[hidden email]> writes:
>> the testsuite hangs in intmath.st. when canceling this the rest of the
>> test suite passes.
>
> Please install GMP and try it then.
> See http://lists.gnu.org/archive/html/help-smalltalk/2008-02/msg00022.html

Nevertheless, I'm starting to suspect that there is a bug somewhere in
the Smalltalk implementation of LargeIntegers...

Paolo


_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk

Ingolf Jandt <[hidden email]> writes:
> I had seen that configure was checking for gmp. But as it did not
> complain I thought it was optional.

It is technically optional.  For example, see the = method in
kernel/LargeInt.st: when GMP is available and used for LargeIntegers,
the primitive (VMpr_LargeInteger_eq, defined in libgst/prims.def) is
used; otherwise, the old representation is used and compared using
Smalltalk code.

GMP only serves to accelerate LargeIntegers; obviously, the effect can
be extreme in certain cases :)

> I will report the result of installing gmp.

One thing I am curious about -- is GMP not available by default on
Darwin/OS X?

--
But you know how reluctant paranormal phenomena are to reveal
themselves when skeptics are present. --Robert Sheaffer, SkI 9/2003


_______________________________________________
help-smalltalk mailing list
[hidden email]
http://lists.gnu.org/mailman/listinfo/help-smalltalk