Thread: [Numpy-discussion] uniform() regression(?) in svn

A package for scientific computing with Python

Brought to you by: charris208, jarrodmillman, kern, rgommers, teoliphant

numpy-discussion

[Numpy-discussion] uniform() regression(?) in svn

From: Andrew J. <a.h...@gm...> - 2006-07-26 13:53:40

Hi-

On PPC Mac OSX universal build 2.4.3, gcc 4.0,


In [1]: import numpy as N

In [2]: print N.__version__
1.0.2897

In [3]: N.random.uniform(0,1)
Segmentation fault

(This originally showed up in the Ticket 83 regression test during 
numpy.test()...)

Andrew

Re: [Numpy-discussion] uniform() regression(?) in svn

From: Travis O. <oli...@ie...> - 2006-07-26 19:14:29

Andrew Jaffe wrote:
> Hi-
>
> On PPC Mac OSX universal build 2.4.3, gcc 4.0,
>
>
> In [1]: import numpy as N
>
> In [2]: print N.__version__
> 1.0.2897
>
> In [3]: N.random.uniform(0,1)
> Segmentation fault
>
> (This originally showed up in the Ticket 83 regression test during 
> numpy.test()...)
>   

This should be O.K. now. 

-Travis

Re: [Numpy-discussion] uniform() regression(?) in svn

From: Darren D. <dd...@co...> - 2006-07-28 18:33:02

On Wednesday 26 July 2006 13:22, Travis Oliphant wrote:
> Andrew Jaffe wrote:
> > Hi-
> >
> > On PPC Mac OSX universal build 2.4.3, gcc 4.0,
> >
> >
> > In [1]: import numpy as N
> >
> > In [2]: print N.__version__
> > 1.0.2897
> >
> > In [3]: N.random.uniform(0,1)
> > Segmentation fault
> >
> > (This originally showed up in the Ticket 83 regression test during
> > numpy.test()...)
>
> This should be O.K. now.

I dont know if this is related or not, but I just did an update from the trunk 
('1.0b2.dev2918') and discovered the following error:

======================================================================
FAIL: Ticket #99
----------------------------------------------------------------------
Traceback (most recent call last):
  
File "/usr/lib64/python2.4/site-packages/numpy/core/tests/test_regression.py", 
line 182, in check_intp
    self.failUnlessRaises(OverflowError,N.intp,'0xb72a7008',16)
AssertionError: OverflowError not raised

----------------------------------------------------------------------

I'm compiling with gcc-4.1.1, python-2.4.3.

Darren

Re: [Numpy-discussion] uniform() regression(?) in svn

From: Stefan v. d. W. <st...@su...> - 2006-07-31 07:51:31

Hi Darren

On Fri, Jul 28, 2006 at 02:33:17PM -0400, Darren Dale wrote:
> I dont know if this is related or not, but I just did an update from th=
e trunk=20
> ('1.0b2.dev2918') and discovered the following error:
>=20
> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
> FAIL: Ticket #99
> ----------------------------------------------------------------------
> Traceback (most recent call last):
>  =20
> File "/usr/lib64/python2.4/site-packages/numpy/core/tests/test_regressi=
on.py",=20
> line 182, in check_intp
>     self.failUnlessRaises(OverflowError,N.intp,'0xb72a7008',16)
> AssertionError: OverflowError not raised

Are you running on a 64 bit machine?  I'll fix the test (which was
written on a 32 bit machine), but I also noticed some strange
behaviour:

In [11]: N.intp('0xfffffffffffffff',16)
Out[11]: 1152921504606846975

In [12]: N.intp('0xffffffffffffffff',16)
-------------------------------------------------------------------------=
--
exceptions.TypeError                                 Traceback (most rece=
nt call last)

/home/stefan/<ipython console>

TypeError: function takes exactly 1 argument (2 given)

Cheers
St=E9fan

Re: [Numpy-discussion] uniform() regression(?) in svn

From: Albert S. <fu...@gm...> - 2006-07-31 11:48:22

Hello all

> <snip>
> Are you running on a 64 bit machine?  I'll fix the test (which was
> written on a 32 bit machine), but I also noticed some strange
> behaviour:
> 
> In [11]: N.intp('0xfffffffffffffff',16)
> Out[11]: 1152921504606846975
> 
> In [12]: N.intp('0xffffffffffffffff',16)
> --------------------------------------------------------------------------
> -
> exceptions.TypeError                                 Traceback (most
> recent call last)
> 
> /home/stefan/<ipython console>
> 
> TypeError: function takes exactly 1 argument (2 given)

I think this ticket might have some pertinent information:

http://projects.scipy.org/scipy/numpy/ticket/99

Cheers,

Albert

Re: [Numpy-discussion] uniform() regression(?) in svn

From: Travis O. <oli...@ie...> - 2006-07-31 16:41:50

Darren Dale wrote:
> On Wednesday 26 July 2006 13:22, Travis Oliphant wrote:
>   
>> Andrew Jaffe wrote:
>>     
>>> Hi-
>>>
>>> On PPC Mac OSX universal build 2.4.3, gcc 4.0,
>>>
>>>
>>> In [1]: import numpy as N
>>>
>>> In [2]: print N.__version__
>>> 1.0.2897
>>>
>>> In [3]: N.random.uniform(0,1)
>>> Segmentation fault
>>>
>>> (This originally showed up in the Ticket 83 regression test during
>>> numpy.test()...)
>>>       
>> This should be O.K. now.
>>     
>
> I dont know if this is related or not, but I just did an update from the trunk 
> ('1.0b2.dev2918') and discovered the following error:
>
> ======================================================================
> FAIL: Ticket #99
> ----------------------------------------------------------------------
> Traceback (most recent call last):
>   
> File "/usr/lib64/python2.4/site-packages/numpy/core/tests/test_regression.py", 
> line 182, in check_intp
>     self.failUnlessRaises(OverflowError,N.intp,'0xb72a7008',16)
> AssertionError: OverflowError not raised
>
> ----------------------------------------------------------------------
>   

No, this is a "test-bug"  This test should only be run on 32-bit systems.

-Travis

Re: [Numpy-discussion] uniform() regression(?) in svn

From: Robert K. <rob...@gm...> - 2006-07-27 14:05:29

Andrew Jaffe wrote:
> Hi-
> 
> On PPC Mac OSX universal build 2.4.3, gcc 4.0,
> 
> 
> In [1]: import numpy as N
> 
> In [2]: print N.__version__
> 1.0.2897
> 
> In [3]: N.random.uniform(0,1)
> Segmentation fault

Travis recently (r2892) checked in a pretty major change to the distributions to 
allow them to broadcast over their arguments. That's probably the source of the 
regression.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless enigma
  that is made terrible by our own mad attempt to interpret it as though it had
  an underlying truth."
   -- Umberto Eco

Re: [Numpy-discussion] uniform() regression(?) in svn

From: Darren D. <dd...@co...> - 2006-07-31 12:36:36

Hi Stefan,

On Monday 31 July 2006 03:51, Stefan van der Walt wrote:
> On Fri, Jul 28, 2006 at 02:33:17PM -0400, Darren Dale wrote:
> > I dont know if this is related or not, but I just did an update from the
> > trunk ('1.0b2.dev2918') and discovered the following error:
> >
> > ======================================================================
> > FAIL: Ticket #99
> > ----------------------------------------------------------------------
> > Traceback (most recent call last):
> >
> > File
> > "/usr/lib64/python2.4/site-packages/numpy/core/tests/test_regression.py",
> > line 182, in check_intp
> >     self.failUnlessRaises(OverflowError,N.intp,'0xb72a7008',16)
> > AssertionError: OverflowError not raised
>
> Are you running on a 64 bit machine?  

Yes, I am.

> I'll fix the test (which was written on a 32 bit machine), 

I just updated from svn; test_regression looks good. Unfortunately, I need to 
report a new error:

======================================================================
ERROR: check_singleton (numpy.lib.tests.test_getlimits.test_longdouble)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/usr/lib64/python2.4/site-packages/numpy/lib/tests/test_getlimits.py", 
line 33, in check_singleton
    ftype = finfo(longdouble)
  File "/usr/lib64/python2.4/site-packages/numpy/lib/getlimits.py", line 49, 
in __new__
    obj = object.__new__(cls)._init(dtype)
  File "/usr/lib64/python2.4/site-packages/numpy/lib/getlimits.py", line 75, 
in _init
    'numpy %s precision floating point number' % precname)
  File "/usr/lib64/python2.4/site-packages/numpy/lib/machar.py", line 210, in 
__init__
    raise RuntimeError
RuntimeError

----------------------------------------------------------------------


> but I also noticed some strange 
> behaviour:
>
> In [11]: N.intp('0xfffffffffffffff',16)
> Out[11]: 1152921504606846975
>
> In [12]: N.intp('0xffffffffffffffff',16)
> ---------------------------------------------------------------------------
> exceptions.TypeError                                 Traceback (most recent
> call last)
>
> /home/stefan/<ipython console>
>
> TypeError: function takes exactly 1 argument (2 given)

Here's what I get:

numpy.intp('0xffffffffffffffff',16)
---------------------------------------------------------------------------
exceptions.OverflowError                             Traceback (most recent 
call last)

/home/darren/<ipython console>

OverflowError: long int too large to convert to int

Darren

Re: [Numpy-discussion] uniform() regression(?) in svn

From: Andrew J. <a.h...@ba...> - 2006-07-31 14:00:30

Hi-

Darren Dale wrote:
> I just updated from svn; test_regression looks good. Unfortunately, I need to 
> report a new error:
> 
> ======================================================================
> ERROR: check_singleton (numpy.lib.tests.test_getlimits.test_longdouble)
> ----------------------------------------------------------------------
> Traceback (most recent call last):
>   File "/usr/lib64/python2.4/site-packages/numpy/lib/tests/test_getlimits.py", 
> line 33, in check_singleton
>     ftype = finfo(longdouble)
>   File "/usr/lib64/python2.4/site-packages/numpy/lib/getlimits.py", line 49, 
> in __new__
>     obj = object.__new__(cls)._init(dtype)
>   File "/usr/lib64/python2.4/site-packages/numpy/lib/getlimits.py", line 75, 
> in _init
>     'numpy %s precision floating point number' % precname)
>   File "/usr/lib64/python2.4/site-packages/numpy/lib/machar.py", line 210, in 
> __init__
>     raise RuntimeError
> RuntimeError


I've found this error, too, under PPC OSX w/gcc 4.0 -- a very different
system, as far as I can tell (Ticket #209). Note that this appears only 
with a > 1 argument to numpy.test(), so not everyone will notice it.