[Stackless] Stackless WSGI server
mprovost at termcap.net
Tue Jul 17 13:25:22 CEST 2007
On Fri, Jul 13, 2007 at 09:05:17AM +0000, Arnar Birgisson wrote:
> On 7/12/07, Christian Tismer <tismer at stackless.com> wrote:
> > I think we don't need to discuss what is possible.
> > Everything is possible, in almost every language.
> > It just does not happen without support.
> Absolutely, sorry for the noise.
> If you guys want to discuss how we would implement a Seaside-like
> framework in Stackless, I'm in. Let me know if this is not the right
> place for that.
> There are some issues (I have to admit I'm not very familiar with
> Seaside though and how/if it solves those issues):
> Sessionless in it's current form breaks the back-button badly. Can we
> solve that somehow while making sense? Would it be feasible (and/or
> desireable) to keep around a serialized version of the tasklets
> several states back? I.e. say I press the back button then the web
> application reverts all of it's state to a previous one by loading an
> "old" version of the tasklet.
> Backing up from state 4 to state 3, and moving from there to a new
> state 4' - should that invalidate state 4 somehow?
> What is the most efficient way to store serialized tasklets, files,
> database, memcached?
> Do we have any way of talking to databases such as sqlite and/or mysql
> and doing file I/O without blocking and without resorting to threads?
Using stackless to pickle sessions with tasklets in them to disk works
great. You only need to get rid of old ones when you run out of disk or
after some reasonable amount of time. The sessions become more like a
tree - so you start with session 1, go to session 2, then if you paste
that URL into a new browser, that creates session 4 in the first
browser and 5 in the second, etc.
Files work for me for storing sessions. As long as you create a new one
on each page they're read-only so you don't need a database to do
locking between processes. (I do use a database to generate the unique
session IDs though.)
It would be great to have something that turned lower level disk I/O
into a channel but that's a whole separate issue.
Stackless mailing list
Stackless at stackless.com
More information about the Stackless