Re: [Pagekit-users] disabling use of session?

[Boris Z. <bo...@2b...>]

Hi Shimon,

Am Freitag, 17. Dezember 2004 16:20 hast Du geschrieben:
> On Fri, Dec 17, 2004 at 12:58:50PM +0100, Boris Zentner wrote:
> > The session's are a tricky thing. But are you really sure that you did
> > not use sessions at all?
> >
> > Off Hand, PageKit does not create a session unless you use the session.
> > If you are logged in, the user  must have a session, since at least the
> > last_activity_time is stored in the session.
>
> In one site, I don't have user logins, and don't call $model->session ever.
> So, I'm pretty sure I never use sessions.  In another site, I have logins

There you should not see any sessions.

> but don't store any data in $model->session.  I don't make use of the
> last_activity_time there - none of the pages or sections require a recent
> login.

But you could use it on a page, pagekit can not know and store the time into 
the session.

( perhaps we can store the last_activity_time in the pkit_user table. And/Or 
make the last_activity thing a config option. )

>
> > > - Can I manually disable instantiation of session if I don't use it? 
> > > (Or is pagekit using it internally for some reason?)
> > >
> > >
> > > - Could PageKit be made a little smarter in order to eliminate my
> > > reason for considering that option?
> >
> > try
> >
> > sub pkit_session_setup {
> >   {}
> > }
> >
> > in Common.pm to trash the session. This might give lots of warnings, but
> > perhaps it works. Please let me know.
>
> This gave me the following warning in the apache error log:
>
> failed to set up session - session_store_class and session_lock_class must
> be defined at /usr/lib/perl5/site_perl/5.8.3/Apache/PageKit.pm line 1258.
>

Thats ok, just comment out the warning and mensure the time four your 
requests. Is there a large speedup now?

> This warning was printed twice per page request - I'm not sure why (there
> are at least three PKIT_COMPONENTs, but only one HTTP request, involved in
> the pages I tried, if that's useful).
>

I take a look, why you see two calls to pkit_session_setup. 

> For additional info, I added a Carp::confess to that the subroutine that
> contains line 1258 of PageKit.pm.  The output is below.  I don't have time
> to trace this call stack now, but perhaps it will be useful to Boris in
> understanding why the session setup got called although I didn't use the
> session.

It is ok to call the pkit_session_setup, if you are a logged user. With 
pkit_session_setup returning a empty hash, we make the session only valid for 
this request and it is not slower than using a hash in perl.

Please mensure if sub pkit_session_setup { {} } is a large speedup. And Ignore 
or comment out the warning. 

>
> thanks,
> shimon.
>
> [Fri Dec 17 10:16:27 2004] [error] [client 66.249.64.197] setup_session at
> /usr/lib/perl5/site_perl/5.8.3/Apache/PageKit.pm line
> 1253\n\tApache::PageKit::setup_session('Apache::PageKit=HASH(0xa59dc34)','u
>ndef')
>
> called at /usr/lib/perl5/site_perl/5.8.3/Apache/PageKit.pm line
> 529\n\tApache::PageKit::prepare_page('Apache::PageKit=HASH(0xa59dc34)')
>
> called at /usr/lib/perl5/site_perl/5.8.3/Apache/PageKit.pm line 216\n\teval
> {...}
>
> called at /usr/lib/perl5/site_perl/5.8.3/Apache/PageKit.pm line
> 210\n\tApache::PageKit::handler('Apache::PageKit','Apache::RequestRec=SCALA
>R(0xa59dd60)')
>
> called at -e line 0\n\teval {...} called at -e line 0\n

-- 
Boris