Re: [ctypes-users] Unit test failures on Mac OS X

[Bob I. <bo...@re...>]

On Apr 28, 2004, at 11:32 PM, Myers Carpenter wrote:

> I've gotten the same results with OS X 10.2, dlcompat, libffi snapshot  
> on the files page of the sf.net project, python 2.3.3 from
>
> http://homepages.cwi.nl/~jack/macpython/macpython-osx.html
>
> And also 10.3 with libffi snapshot on the files page of the sf.net  
> project (dlcompat included with 10.3, using stock 10.3 python).
>
> I will be doing some more investigation, (the first error looks like a  
> bug in the test code), but I thought I'd let everyone know.
>
> myers
>
>
> $ python setup.py test
> running test
> running build
> running build_py
> running build_ext
> ...F..........FF...................................................E... 
> ............
> ======================================================================
> ERROR: test_functions (test_stringptr.StringPtrTestCase)
> ----------------------------------------------------------------------
> Traceback (most recent call last):
>   File "unittests/test_stringptr.py", line 44, in test_functions
>     clib = CDLL("/lib/libc.so.6")
>   File  
> "build/lib.darwin-7.3.0-Power_Macintosh-2.3/ctypes/__init__.py", line  
> 222, in __init__
>     self._handle = LoadLibrary(self._name)
> OSError: dlcompat: Can not open "/lib/libc.so.6"

The equivalent on OS X is "/usr/lib/libc.dylib".  The test is more or  
less hardcoded for a particular version of Linux.  The test works if  
it's fixed (not in patch).

> ======================================================================
> FAIL: test_integrate (test_callbacks.CallbacksTestCase)
> ----------------------------------------------------------------------
> Traceback (most recent call last):
>   File "unittests/test_callbacks.py", line 24, in test_integrate
>     self.failUnless(diff < 0.01)
>   File  
> "/System/Library/Frameworks/Python.framework/Versions/2.3/lib/ 
> python2.3/unittest.py", line 268, in failUnless
>     if not expr: raise self.failureException, msg
> AssertionError

This is a legitimate problem, the func(x) is always receiving some  
infinitesimal number across the bridge and and returning 0.00 as far as  
C is concerned.  It seems that ctypes is just broken, it's passing  
ffi_type_sint as all the parameters for callbacks.  I decided to fix  
this, see attached patch.  Not the best solution, but the least amount  
of code.

> ======================================================================
> FAIL: test_intresult (test_functions.FunctionTestCase)
> ----------------------------------------------------------------------
> Traceback (most recent call last):
>   File "unittests/test_functions.py", line 106, in test_intresult
>     self.failUnless(result == 21)
>   File  
> "/System/Library/Frameworks/Python.framework/Versions/2.3/lib/ 
> python2.3/unittest.py", line 268, in failUnless
>     if not expr: raise self.failureException, msg
> AssertionError
>
> ======================================================================
> FAIL: test_longlong_callbacks (test_functions.FunctionTestCase)
> ----------------------------------------------------------------------
> Traceback (most recent call last):
>   File "unittests/test_functions.py", line 300, in  
> test_longlong_callbacks
>     self.failUnless(13577625587 == f(1000000000000, cb))
>   File  
> "/System/Library/Frameworks/Python.framework/Versions/2.3/lib/ 
> python2.3/unittest.py", line 268, in failUnless
>     if not expr: raise self.failureException, msg
> AssertionError

These two are almost certainly the same class of error as the callbacks  
were..  the ctypes ffi stuff isn't in a finished state.  I'll take a  
look later this week.