It's hard to sort out what the bug was, exactly. So, Big Hammer:
1. Python shouldn't be in the business of #define'ing NULL, period.
2. Users of the Python C API shouldn't be in the business of not including
Python.h, period.
Hence:
1. Removed all #define's of NULL in Python source code (pyport.h and
object.h).
2. Since we're *relying* on stdio.h defining NULL, put an #error in
Python.h after its #include of stdio.h if NULL isn't defined then.
comments, docstrings or error messages. I fixed two minor things in
test_winreg.py ("didn't" -> "Didn't" and "Didnt" -> "Didn't").
There is a minor style issue involved: Guido seems to have preferred English
grammar (behaviour, honour) in a couple places. This patch changes that to
American, which is the more prominent style in the source. I prefer English
myself, so if English is preferred, I'd be happy to supply a patch myself ;)
Stein -- thanks!). Incidentally removed all the Py_PROTO macros
from object.h, as they prevented my editor from magically finding
the definitions of the "coercion", "cmpfunc" and "reprfunc"
typedefs that were being redundantly applied in longobject.c.
errors in some of the hash algorithms. For exmaple, in float_hash and
complex_hash a certain part of the value is not included in the hash
calculation. See Tim's, Guido's, and my discussion of this on
python-dev in May under the title "fix float_hash and complex_hash for
64-bit *nix"
(2) The hash algorithms that use pointers (e.g. func_hash, code_hash)
are universally not correct on Win64 (they assume that sizeof(long) ==
sizeof(void*))
As well, this patch significantly cleans up the hash code. It adds the
two function _Py_HashDouble and _PyHash_VoidPtr that the various
hashing routine are changed to use.
These help maintain the hash function invariant: (a==b) =>
(hash(a)==hash(b))) I have added Lib/test/test_hash.py and
Lib/test/output/test_hash to test this for some cases.
This patch modifies the type structures of objects that
participate in GC. The object's tp_basicsize is increased when
GC is enabled. GC information is prefixed to the object to
maintain binary compatibility. GC objects also define the
tp_flag Py_TPFLAGS_GC.
Improvements:
- does no longer need any extra memory
- has no relationship to tstate
- works in debug mode
- can easily be modified for free threading (hi Greg:)
Side effects:
Trashcan does change the order of object destruction.
Prevending that would be quite an immense effort, as
my attempts have shown. This version works always
the same, with debug mode or not. The slightly
changed destruction order should therefore be no problem.
Algorithm:
While the old idea of delaying the destruction of some
obejcts at a certain recursion level was kept, we now
no longer aloocate an object to hold these objects.
The delayed objects are instead chained together
via their ob_type field. The type is encoded via
ob_refcnt. When it comes to the destruction of the
chain of waiting objects, the topmost object is popped
off the chain and revived with type and refcount 1,
then it gets a normal Py_DECREF.
I am confident that this solution is near optimum
for minimizing side effects and code bloat.
/* More standard operations (at end for binary compatibility) */
should now be:
/* More standard operations (here for binary compatibility) */
since they're no longer at the end!
Added wrapping macros to dictobject.c, listobject.c, tupleobject.c,
frameobject.c, traceback.c that safely prevends core dumps
on stack overflow. Macros and functions in object.c, object.h.
The method is an "elevator destructor" that turns cascading
deletes into tail recursive behavior when some limit is hit.
a new proc type (objobjproc), a new slot sq_contains to
PySequenceMethods, and a new flag Py_TPFLAGS_HAVE_SEQUENCE_IN to
Py_TPFLAGS_DEFAULT. More to follow.
Add a new member to the PyBufferProcs struct, bf_getcharbuffer. For
backward compatibility, this member should only be used (this includes
testing for NULL!) when the flag Py_TPFLAGS_HAVE_GETCHARBUFFER is set
in the type structure, below. Note that if its flag is not set, we
may be looking at an extension module compiled for 1.5.1, which will
have garbage at the bf_getcharbuffer member (because the struct wasn't
as long then). If the flag is one, the pointer may still be NULL.
The function found at this member is used in a similar manner as
bf_getreadbuffer, but it is known to point to 8-bit character data.
(See discussion in getargs.c checked in later.)
As a general feature for extending the type structure and the various
structures that (may) hang off it in a backwards compatible way, we
rename the tp_xxx4 "spare" slot to tp_flags. In 1.5.1 and before,
this slot was always zero. In 1.5.1, it may contain various flags
indicating extra fields that weren't present in 1.5.1. The only flag
defined so far is for the bf_getcharbuffer member of the PyBufferProcs
struct.
Note that the new spares (tp_xxx5 - tp_xxx8), once they become used,
should also be protected by a flag (or flags) in tp_flags.
Py_DECREF, to reduce the warnings when compiling with reference count
debugging on. (There are still warnings for each call to
_Py_NewReference -- too bad.)
use the new names exclusively, and the linker will see the new names.
Files that import "Python.h" also only see the new names. Files that
import "allobjects.h" will continue to be able to use the old names,
due to the inclusion (in allobjects.h) of "rename2.h".
object.h: made sizes and refcnts signed ints.
stringobject.h: make getstrsize() signed int.
methodobject.h: add METH_VARARGS and METH_FREENAME flag bit definitions.
Tim Peters
Guido van Rossum
Thomas Wouters
Fred Drake
Greg Stein
Jeremy Hylton
Sjoerd Mullender
Jack Jansen