Re: Portlet performance

2006-06-13 Thread John Singleton
I looked into the issue with the start up time for Tapestry Portlets I was 
having, and I added a JIRA issue for it. It seems like the initialisation of 
the portlet is relatively short, but the initial render takes many seconds. 
This is proably due to lazy initialisation meaning that a lot of the work is 
delayed until the initial render?

I'm now in the situation where I hava  JSF and a Tapestry version of a portlet, 
both of which do the same thing but the JSF version can be added to the portal 
in less than a second, the Tapestry version takes about 10. It's looking like I 
am going to have to go down the JSF/Facelets path rather than Tapestry. I 
prefer the Tapestyr aproach, but we can't take the performance hit.

Is there anything I can provide to help find the real source of the problem, or 
is it unlikely to be fixable?

Thanks

John.



-
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]



Re: Portlet performance

2006-06-07 Thread John Singleton


Jesse Kuhnert jkuhnert at gmail.com writes:

 
 First I've heard of it. I didn't initially have plans on doing anything with
 it but I'd be happy to in order to save you from jsf. :)
 
 Whatever it is probably won't happen in the initial alpha release as I
 have some other items taking higher priority, but if you start a jira issue
 with some clear parameters for me to test with we can probably figure
 something out.
 
I've done a few more timings, and it seems like the Registry construction isn't
the culprit.

ApplicationPortlet.init takes about 2.5s (.6 of which is the constructRegistry
call). The initial render takes 4.5s . Will this initial hit be when it
constructs the State Object that I am injecting?

I'll create a Jira issue, but just wanted to make sure I wasn't doing something
silly first.

Thanks

John


-
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]



Portlet page resolving.

2006-05-30 Thread John Singleton
I've written my own PortletPageResolver, to resolve portlet pages to subdirs -
i.e. View gets resolved to portlet-name/View 

This works fine, except when I want to navigate to a different make. When I
submit a form I want to go to a page called Winner. From the form success
listener I've tried returning Winner, injecting the Winner into the View and
returning the IPage. Both these resulted in Page 'Winner' not found in
application namespace.


The only thing that has worked so far is returning portlet-name/Winner. 

It seems like when I return Winner my PortletPageResolver is not being called.
Is it because the PortletPageResolver only resolves the Home page location? 

Thanks

John


-
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]



Portlet performance

2006-05-26 Thread John Singleton
I'm evaluating Tapestry, and facelets, for use in a Portlet based project. We
need to migrate an exsiting Servlet/JSP based application to Portlets.

Currently I'm leaning towards Tapestry, from a coding standpoint it seems much
cleaner to me. The only problem is the initial overhead when adding a portlet to
a page. I understand this is because creating the per Portlet instance of the
Hivemind registry takes some time. My simple test portlet written in facelets
adds to the page in about a second, but the tapestry version takes about 10. The
customer probably won't stand for this.

Is there anything I can do to reduce this initial hit? Is this something being
addressed in 4.1 ?

Thanks

John


-
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]