On Wed, 17 Mar 2010 03:07:57 +, Michael Rynn wrote:
On Tue, 16 Mar 2010 16:12:25 +, Moritz Warning wrote:
On Tue, 16 Mar 2010 04:41:55 +, Michael Rynn wrote:
[..]
But do not take this very restricted and rather extreme sampling of
possible test parameters to be a reliable
On Wed, 17 Mar 2010 02:57:34 -0300, Michael Rynn
michaelr...@optusnet.com.au wrote:
[snip]
I want to add that the problems of AA with garbage collection are not
particular to the implementation of built-in AA. The number of nodes that
builtin AA creates will increase GC scanning, but the AA
On Tue, 16 Mar 2010 16:12:25 +, Moritz Warning wrote:
On Tue, 16 Mar 2010 04:41:55 +, Michael Rynn wrote:
The use of built in Associative Array can have deleterious effects on
Garbage Collection.
:(
Fwiw, here is the original Python Dictionary implementation:
I've been arguing for a long time that the interaction between conservative GC
w/o
thread-local allocators and the builtin AA is horrible. Keep this in mind: You
have to take the GC lock for **EVERY SINGLE INSERTION** into a builtin AA. This
makes them absolutely worthless in multithreaded
On Tue, 16 Mar 2010 04:41:55 +, Michael Rynn wrote:
The use of built in Associative Array can have deleterious effects on
Garbage Collection.
:(
Fwiw, here is the original Python Dictionary implementation:
http://svn.python.org/projects/python/trunk/Objects/dictobject.c
This is also a
On Tue, 16 Mar 2010 16:12:25 +, Moritz Warning wrote:
On Tue, 16 Mar 2010 04:41:55 +, Michael Rynn wrote:
The use of built in Associative Array can have deleterious effects on
Garbage Collection.
:(
Fwiw, here is the original Python Dictionary implementation:
The use of built in Associative Array can have deleterious effects on
Garbage Collection.
D by default uses garbage collection for memory management.
This means that any blocks of memory that contain random bits can be
mistaken as containing pointers to valid data, and can prevent complete