[Python-Dev] Re: [Python-checkins]python/dist/src/Modules

gcmodule.c,2.33.6.5,2.33.6.6 (original) (raw)

Tim Peters tim.one@comcast.net
Wed, 09 Apr 2003 22:29:21 -0400

[Greg Ewing]

Failing that, perhaps they should be made a bit less dynamic, so that the GC can make reasonable assumptions about their existence without having to execute Python code.

[Tim]

Guido already did so for new-style classes in Python 2.3. That machinery doesn't exist in 2.2.2, and old-style classes remain a problem under 2.3 too. Backward compatibility constrains how much we can get away with, of course.

FYI, those who study the checkin comments know how this ended. It ended well! gc no longer does anything except string-keyed dict lookups when determining whether a finalizer exists, for old- & new- style classes, and in 2.3 CVS & the 2.2 maintenance branch.

The only incompatibilities appear to be genuine bug fixes. The hasattr() method was actually incorrect in two mondo obscure cases (one where hasattr said "yes, del exists" when a finalizer couldn't actually be run, and the other where hasattr said "no, del doesn't exist" when arbitrary Python code actually could be invoked by destructing an object). A new private API function _PyInstance_Lookup was added in 2.2 and 2.3, which does for old-style class instances what _PyType_Lookup does for new-style classes (determines whether an attribute exists via pure C string-keyed dict lookups).