On Feb 17, 5:24 pm, mdipierro <mdipie...@cs.depaul.edu> wrote:
> You are right this not a reason not to move. The main reason is
> backward compatibility. In general 3.x seems to be slower than 2.x for
> data intensive processes. Am I wrong?

The Python 3.X stream is getting better performance wise and the GIL
changes they are working on that that bug is finding an issue with,
promise to certainly yield better performance for a threaded system on
a multi processor or multi core system. This could be significant for
threaded WSGI systems which in particular could benefit from it
greatly. Would certainly be better than Python 2.X given the issues
with it as described in that PDF analysis.

That performance is still not as good is not really a good reason to
delay starting to move a code base to Python 3.X. After all, it would
likely take 6 months or more to move a large web framework across and
be confident that the changes to the framework are all correct. By
then the performance of Python 3.X stream could have improved to make
it more usable.

Graham

> On Feb 17, 12:05 am, Graham Dumpleton <graham.dumple...@gmail.com>
> wrote:
>
>
>
> > On Feb 17, 4:20 pm, mdipierro <mdipie...@cs.depaul.edu> wrote:
>
> > > Look at this bug report
>
> > >http://bugs.python.org/issue7946
>
> > > specifically look at the benchmarks at the bottom. This kind of bugs
> > > affects web applications since they have threads with lots of IO.  One
> > > more reason to stay with python 2.5 for now.
>
> > For some background context, read:
>
> >    http://www.dabeaz.com/python/NewGIL.pdf
>
> > Based on the name, same guy as reported the bug has previously worked
> > in getting the GIL behaviour changed in Python 3.2 to address issues
> > in linked PDF. That bug report seems just to say that the changes they
> > made in Python 3.2 so far are not perfect as is showing up this bad
> > degradation for a particular use case. Just means they need to address
> > that particular behaviour and further optimise the code behaviour to
> > counter it before they release Python 3.2.
>
> > Overall the changes made to Python 3.2 GIL seem to be positive. Once
> > they sort any issues like this out, it would be good to see it back
> > ported to Python 2.X stream.
>
> > Anyway, not sure why you are holding this up as reason to stay with
> > Python 2.5. Python 2.6 and 3.1 both have old GIL code still and Python
> > 3.2 isn't even released yet, so this is a bug for an unreleased
> > version.
>
> > Graham

-- 
You received this message because you are subscribed to the Google Groups 
"web2py-users" group.
To post to this group, send email to web...@googlegroups.com.
To unsubscribe from this group, send email to 
web2py+unsubscr...@googlegroups.com.
For more options, visit this group at 
http://groups.google.com/group/web2py?hl=en.

Reply via email to