[Patches] [ python-Patches-486743 ] remove bad INCREF, propagate exception

[noreply@sourceforge.net]

Patches item #486743, was opened at 2001-11-28 15:11
You can respond by visiting: 
http://sourceforge.net/tracker/?func=detail&atid=305470&aid=486743&group_id=5470

Category: None
Group: None
Status: Closed
Resolution: Accepted
Priority: 5
Submitted By: Zooko Ozoko (zooko)
Assigned to: Martin v. Löwis (loewis)
Summary: remove bad INCREF, propagate exception

Initial Comment:
This is my very first foray into the Python
interpreter, so please be patient.  It looks like to me
that this INCREF is wrong since the PyList_Append()
will steal the reference.  Also I tested it.  Current
CVS does this:

>>> import gc
>>> len(gc.get_objects())
1559
>>> len(gc.get_objects())
1562
>>> len(gc.get_objects())
1565
>>> len(gc.get_objects())
1568
>>> len(gc.get_objects())
1571
>>> len(gc.get_objects())
1574

and with my patch it does this:

>>> import gc
>>> len(gc.get_objects())
1559
>>> len(gc.get_objects())
1559
>>> len(gc.get_objects())
1559
>>> len(gc.get_objects())
1559
>>> len(gc.get_objects())
1559
>>> len(gc.get_objects())
1559
>>> len(gc.get_objects())
1559


In addition I added checks for error return from
PyList_Append(), and propagated the exception condition
up the callstack.


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

>Comment By: Zooko Ozoko (zooko)
Date: 2001-12-02 10:12

Message:
Logged In: YES 
user_id=52562

Whoo-hoo!  My patch was accepted!  :-)

Here's the remaining diffs between my sandbox and the
current CVS.  Not sure if there is something wrong with this
or if you just didn't see it:

(The same thing is also attached, since I guess my web
browser will mangle it.)


@@ -724,6 +758,9 @@
        if (!PyArg_ParseTuple(args, ":get_objects")) /*
check no args */
                return NULL;
        result = PyList_New(0);
+       if (result == NULL) {
+               return NULL;
+       }
        if (append_objects(result, &_PyGC_generation0) ||
            append_objects(result, &generation1) ||
            append_objects(result, &generation2)) {



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

Comment By: Martin v. Löwis (loewis)
Date: 2001-12-02 04:22

Message:
Logged In: YES 
user_id=21627

Thanks for the patch, committed as gcmodule.c 2.32 (with
slight modifications: return 0 in case of success).

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

Comment By: Zooko Ozoko (zooko)
Date: 2001-11-28 16:33

Message:
Logged In: YES 
user_id=52562

Whoops.  It is against dist/src/Modules/gcmodule.c

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

Comment By: Gregory P. Smith (greg)
Date: 2001-11-28 16:30

Message:
Logged In: YES 
user_id=413

which file is the patch against?

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

Comment By: Zooko Ozoko (zooko)
Date: 2001-11-28 16:12

Message:
Logged In: YES 
user_id=52562

Oh, and here is a new version of the patch, which also tests
for failure from PyList_New().



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

You can respond by visiting: 
http://sourceforge.net/tracker/?func=detail&atid=305470&aid=486743&group_id=5470