[pypy-issue] [issue1122] [cpyext] PyVarObject builtins and definition of Py_SIZE()

Stefan Behnel tracker at bugs.pypy.org
Tue Apr 10 22:50:51 CEST 2012


New submission from Stefan Behnel <stefan_ml at behnel.de>:

I wonder why the definition of PyTupleObject and friends use PyObject_HEAD
instead of PyObject_VAR_HEAD - how is Py_SIZE(atuple) supposed to work?
Py_SIZE() is currently defined like this:

       #define Py_SIZE(ob)             (((PyVarObject*)(ob))->ob_size)

i.e. it actually requires a PyVarObject struct. That macro works with quite a
number of builtin types in CPython, but it can't work for them in PyPy.

Personally, I wouldn't mind adding the ob_size field to at least the constant
sequence types as a duplication of the object length value, because it would
allow for a similar optimisation of the couple of Py(Tuple|Bytes|Unicode|maybe
others)_GET_SIZE() macros as for the ones in issue1121, by simply aliasing them
to Py_SIZE(). C extensions commonly expect these macros to be fast.

----------
messages: 4221
nosy: pypy-issue, sbehnel
priority: bug
status: unread
title: [cpyext] PyVarObject builtins and definition of Py_SIZE()

________________________________________
PyPy bug tracker <tracker at bugs.pypy.org>
<https://bugs.pypy.org/issue1122>
________________________________________


More information about the pypy-issue mailing list