Webware for Python

On Friday 07 December 2001 11:08 am, Geoffrey Talvola wrote:
> I'm taking a closer look at the implementation of
> SessionDynamicStore.  It looks like it actually checks the filesystem
> on _every_ request, even if the session is in memory.  This would
> seem to slow it down considerably.
>
> That plus the concurrency issue and it looks like SessionDynamicStore
> could stand a rewrite.

It's other deficiency that I recall is that it's hard coded to 
FileStore. If someone had a SQL store, or something else, they might 
want dynamic to use that instead.

-Chuck

On Friday 07 December 2001 13:12, Chuck Esterbrook wrote:
> On Friday 07 December 2001 11:08 am, Geoffrey Talvola wrote:
> > I'm taking a closer look at the implementation of
> > SessionDynamicStore. =A0It looks like it actually checks the
> > filesystem on _every_ request, even if the session is in memory.
> > =A0This would seem to slow it down considerably.
> >
> > That plus the concurrency issue and it looks like
> > SessionDynamicStore could stand a rewrite.
>
> It's other deficiency that I recall is that it's hard coded to
> FileStore. If someone had a SQL store, or something else, they
> might want dynamic to use that instead.

Have a look at the DynamicSessionStore in the file I've attached. =20
It's just been rewired to handle Chuck's proposal.  It's based on the=20
experimental code, but most of it should transfer across easily

On Friday December 07, 2001 06:46 pm, Tavis Rudd wrote:
> On Friday 07 December 2001 13:12, Chuck Esterbrook wrote:
> > On Friday 07 December 2001 11:08 am, Geoffrey Talvola wrote:
> > > I'm taking a closer look at the implementation of
> > > SessionDynamicStore. =A0It looks like it actually checks the
> > > filesystem on _every_ request, even if the session is in memory.
> > > =A0This would seem to slow it down considerably.
> > >
> > > That plus the concurrency issue and it looks like
> > > SessionDynamicStore could stand a rewrite.
> >
> > It's other deficiency that I recall is that it's hard coded to
> > FileStore. If someone had a SQL store, or something else, they
> > might want dynamic to use that instead.
>
> Have a look at the DynamicSessionStore in the file I've attached.
> It's just been rewired to handle Chuck's proposal.  It's based on the
> experimental code, but most of it should transfer across easily

I took a look.  It doesn't seem to address the thread-safety concerns, an=
d=20
also I honestly don't see the point of using anything other than the File=
=20
store as the secondary store for Dynamic.  The nature of the Dynamic stor=
e=20
prevents different copies of the appserver from sharing sessions anyway, =
so=20
there wouldn't be much point to a SQL or ZODB secondary store...

- Geoff