New submission from Lisandro Dalcin :
I've built Python-2.6.5 form with this :
$ ./configure --without-threads --prefix=$HOME/python-without-threads
$ make && make install
$ cd ~/python-without-threads/lib
$ nm libpython2.6.so | grep PyGILState
$
At Include/pystate.h, PyGIL
New submission from Lisandro Dalcin <[EMAIL PROTECTED]>:
from warnings import warn
warn("hello world") # -> Success
warn(UserWarning) # -> Segmentation fault
warn(None) # -> Segmentation fault
warn(1) # -> Segmentation fault
--
New submission from Lisandro Dalcin :
At Objects/longobject.c, in almost all cases
OverflowError is raised when a unsigned integral is requested from a
negative PyLong. However, this one breaks the rules:
int
_PyLong_AsByteArray(PyLongObject* v,
unsigned char* bytes, size_t n
Lisandro Dalcin added the comment:
I can contribute a patch. However, I would like to wait until Tim
comments on this.
___
Python tracker
<http://bugs.python.org/issue5
Lisandro Dalcin added the comment:
Mark, here you have a patch. I've only 'make test' on a 32bit Linux box
Just two comments:
- in docs: perhaps the 'versionchanged' stuff should be added.
- in tests: I did not touch Modules/_testcapimodule.c, as it seems the
test
Lisandro Dalcin added the comment:
It worked for me.
BTW, 'make test' did not noticed the change in Modules/testcapi_long.h,
which is #include'd by Modules/_testcapimodule.c. I've attached a
trivial patch for setup.py fixing the dependency issue.
Added file: http://bugs.
New submission from Lisandro Dalcin :
structmember.h lacks a 'T_SIZET' define in order to properly support
struct fields of type 'size_t' within PyMemberDef.
If this feature request is accepted, I can provide the patches.
For 2.7, It would be nice to do:
#de
Lisandro Dalcin added the comment:
In 2.6 and 3.0, we already have 'Py_ssize_t', see 'T_PYSSIZET'
definition on Include/structmember.h. I'm asking for 'size_t' support.
___
Python tracker
Lisandro Dalcin added the comment:
Two questions:
1) Why do you prefer to pass destructor as a first argument?
2) Do we really need two different calls for providing a context
pointer? Why no just one call and pass a NULL context?
A comment:
Suppose one wants to implement export/import
Lisandro Dalcin added the comment:
In Doc/c-api/capsule.rst, you wrote
.. cfunction:: int PyCapsule_Import(const char* name, int no_block)
but it should be:
.. cfunction:: void* PyCapsule_Import(const char* name, int no_block)
Additionally, you wrote "disambugate" in m
New submission from Lisandro Dalcin :
When doctests are written in docstrings from C extension modules,
'doctest' reads the binary extension module file.
The attached one-line patch seems to fix the problem, it is in fact very
similar to patch for issue4050 related t
Lisandro Dalcin added the comment:
I've tested latest David's patch against Cython test suite (doctests
living in extension modules), and all is working as expected.
--
___
Python tracker
<http://bugs.python.
12 matches
Mail list logo
