Cool, I'll see about updating the download webpage some time today thereby
unleashing 3.2.10...
On Wed, 2 Aug 2006, Graham Dumpleton wrote:
Grisha wrote ..
Is this a mod_python problem? Should we scrap 3.2.10? - it's out there,
but the download page has not beed updated and no announcement made yet.
It is a mod_python problem. See:
https://issues.apache.org/jira/browse/MODPYTHON-181
It only manifests when mod_python handlers are registered in two distinct
phases for a specific request. It has been there since at least 3.1.X.
No one has probably noticed as just about everyone only bothers to use a
response handler and never in combination with handlers running in other
phases.
Given that it probably affects a small percentage of users and no one has
seemed to notice before now, the feeling was that we didn't want to delay a 3.2
update yet again, especially since other fixes for memory leaks in forms
handling in 3.2.10 are probably more significant.
The memory leak in 3.2 is also probably smaller than what I am seeing in 3.3
because of other changes that have been made in 3.3 which would make the
problem worse. Thus impact of not fixing it in 3.2 right now is less.
We also figured there would be another 3.2 bug rollup release before 3.3 is
released so that we ensure last 3.2 version is as good as it can be before
new features introduced in 3.3.
Graham
On Wed, 2 Aug 2006, Deron Meranda wrote:
On 7/31/06, Graham Dumpleton <[EMAIL PROTECTED]> wrote:
Are you using PythonAccessHandler at same time as PythonHandler?
Aren't you seeing the hideously bad memory leak finally uncovered
in the past day?
Yes I am. With 3.2.10 I am seeing about 750 kB leaked per
about 10,000 requests (on an IA32 arch). I don't know if this
is more or less of a leak than expected. I see you've got it
fixed on the trunk though.
--
Deron Meranda
_______________________________________________
Mod_python mailing list
[EMAIL PROTECTED]
http://mailman.modpython.org/mailman/listinfo/mod_python
