On Wed, Oct 26, 2011 at 4:28 PM, Jim Fulton <j...@zope.com> wrote:
>>> If this is a customer project, I'd recommend dropping either
>>> iron python or ZODB at this stage.
>> Why that?! As I know, ironpython creates with the JIT compiler bytecode
>> in realtime.
> Because this is going to be pretty new code and you'll be a trail blazer.
AFAIK iron python is not considerably faster than cpython, so I think
you will end up with exactly the same performance problems of cpython
using a pure python ZODB on ironpython.
>> what about pypy ?!
> It will be interesting to see how much of the performance pypy
> can gain back.
As has been shown with the json module pypy can be really fast with
pure python code, in the next version of pypy the pure python json
module is faster than the c extension optimizer _json on cpython. So I
really think ZODB can be faster on pypy, the problem being how much
work would it be?
But looking at a future of a pure python ZODB faster than the one we
have today (that is quite fast) would be awesome. Not so much for the
speed but to be able to implement everything in the language we all
Good luck with your work.
For more information about ZODB, see http://zodb.org/
ZODB-Dev mailing list - ZODB-Dev@zope.org