"Kevin A. Burton" wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Alan McAuley <[EMAIL PROTECTED]> writes:
>
> > All,
> >
> > Out of interest, has anyone performed any sustained testing of Jetspeed
> > as a package.
> >
> > Would be interested to hear and share peoples experiences in this area.
>
> Someone did a while back. Check the archives for the actual numbers. Basically
> on a PIII 450 /w I think 128M of RAM and 1000 concurrent simulated users
> Jetspeed could throw around 300 page views per second. That will vary of course
> with your hardware/software config.
>
Was it using jdk1.3 Hotspot server 2.0?
If it was not, then we should expect a noticeable speed increase when this VM
becomes available in all architectures. Currently it is the best solution for
Windows. Even "out of the box" jdk1.3 for Linux should be better than any previous
solution.
In my collaboration with Soluziona people, I will have to stress test it, so I will
report soon current status with "modern" VMs.
>
> > eg load testing, thread safety, functional etc.
> >
> > I did read here once of some load testing that was acceptable to the
> > client but that Kevin Burton thought was poor performance considering
> > the cacheing mecahanisms employed.
> <snip>
>
> well... I am usually too optimistic!
>
I think there are a few remaining thread safety issues regarding PortletConfig, but
they should only do things like mixing property values. I will solve them soon. Even
I'm not sure the property values mixed are being used anywhere. So I think branch
pre-proposal-0003-no-psml should be fairly thread safe by now.
We were discussing issues about thread safety.
We need to clarify different Contexts being used in the layout process:
- No mutable objects should be stored in the Portlet, Controller, etc. Objects that
depend on the user should go to the Session object, while objects depending on the
request whould go to the Request (or RunData) object
- RunData is to be used always from the getContent(RunData data) parameter. We could
extend it or pass a different object, if we respect the same convention.
- JetspeedConfig is safe if we use it as a read only configuration reflecting walues
in the jcfg file
- A new (JetspeedContext?) object can be passed to the layout containing properties
coming from user psml. This object should be kept in the Session, a natural place to
store user psml once scanned.
- A mechanism for scoping of the URIs generated is needed if we want to be able to
use the same portlet twice in a page and to avoid parameter name clashes.
>
> Kevin
> - --
> Kevin A Burton ([EMAIL PROTECTED], [EMAIL PROTECTED], UIN: 73488596)
>
> ~/.signature: No such file or directory
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.0.1 (GNU/Linux)
> Comment: Get my public key at: http://relativity.yi.org/pgpkey.txt
>
> iD8DBQE5+BzDAwM6xb2dfE0RAiyNAKDTQq8hdVizzEYU9MCaIIHN0FEO3wCeNmO9
> DBUrz6w08zrmOsnxF145dK4=
> =tBpY
> -----END PGP SIGNATURE-----
>
> DES spy Peking FSF KGB quiche [Hello to all my fans in domestic surveillance]
> Honduras $400 million in gold bullion Khaddafi BATF radar strategic Nazi CIA
>
> --
> --------------------------------------------------------------
> Please read the FAQ! <http://java.apache.org/faq/>
> To subscribe: [EMAIL PROTECTED]
> To unsubscribe: [EMAIL PROTECTED]
> Archives and Other: <http://java.apache.org/main/mail.html>
> Problems?: [EMAIL PROTECTED]
--
--------------------------------------------------------------
Please read the FAQ! <http://java.apache.org/faq/>
To subscribe: [EMAIL PROTECTED]
To unsubscribe: [EMAIL PROTECTED]
Archives and Other: <http://java.apache.org/main/mail.html>
Problems?: [EMAIL PROTECTED]
