Re: [studs-user] isCancelled() in GolfLog demo webapp

[Dan A. <dan...@gm...>]

> Yes. Actually, after I sent the email, I continued my own investigation,
> and noticed that isCancelled() actually was set to false, but as you
> said, is was in fact isValidToken() that failed because of the failed
> session. The strange thing (for me) was that the session was actually
> started (or at least - a PHPSESSID cookie was set for the correct URL
> and sent to web server on each request). I didn't get any further since
> dinner with my girlfriend was on the schedule (and just when things
> started to get exciting ;).

Got some updates for you.  Turns out, the session problem was actually
an RTFM thing.  The sourceforge docs say that if you want PHP sessions
to work correctly, you have to set the session save path to
/tmp/persistent/<projectname> or else it will be periodically deleted.
 So what was happening was the session was being destroyed every
couple of seconds.  Turns out that issue didn't have anything to do
with Studs.

However, your report has helped to cleanup the handling of the
sessions rather nicely.  I feel very good about their stability now.=20
Unfortunately, the mysql server is misconfigured (or just overloaded)
on sourceforge, so the demo is suffering from that.  I took Studs
aside and did some testing with the mysql connections in the golflog
demo and verified that I never open more than one connection to the
mysql server, no matter how many times I hit Golflog.

> Excellent! Actually, I think I'll just take they time to appreciate what
> you are doing. I've been evaluating Studs and Mojavi, and right now
> Studs is obviously the right choice. It's more similar to Struts (which
> I like) and a lot more active. Mojavi has been in 3.0 development (and
> therefore, awaiting PHP5 compliance) for over 8 months. Anyway - keep up
> the excellent work!

Whooohoooo!!!  Thanks so much!

/dan

--=20
Open Source Advocacy
http://www.mojavelinux.com