Yep, the current portlet dev is really complicated, but there will be a huge work to do if we decide to switch pluto to other framework like JSF... not sure how much for Pluto2.
And I think we don't have enough time for the migration before G2.2 release..
If we're going to rewrite bits of the portal, we should consider moving to pluto 2. IIUC there are a bunch of features in portlet 2 spec that may make our portlets simpler. I also think we should investigate frameworks such as jsf or even wicket or something because the current portlet code is ridiculously complicated for what it does. There must be a more sensible way to write a web app.thanksdavid jencksOn Jul 1, 2009, at 9:41 AM, Joseph Leong wrote:So unfortunately what happened between Dojo 0.4.3-> Mostly anything newer especially 1.3.1 is that they had the idea to classify their libraries to "Dijit" (Widgets) and other subsections. As such, the porting effort is not small. I believe the debug-views portlets and such still depend on 0.4.3. At this point in time, my opinion would be to not try and migrate any 0.4.3 dependent code. There has been so much change between the dojo versions that it would be probably simpler and cleaner to just rewrite these portlets. I think it'd be a good choice to get rid of the old Dojo libraries once and for all as they add a bit to the geronimo footprint size.. not to mention there are a lot more features in the latest Dojo release that can probably accomplish what you wanted to in the older versions.
Joseph LeongOn Wed, Jul 1, 2009 at 12:10 PM, David Jencks <email@example.com> wrote:
Newer would be better if we can get it to work. I set this up a few days ago and forgot the details... I think that I saw some problem and wasn't sure what was causing it and tried changing to an earlier dojo version. I didn't actually have any reason to think the problem was caused by dojo so very likely the more recent release should work.On Jul 1, 2009, at 1:14 AM, Ivan wrote:I think the one is what need, no samples and testcases are included. But I found 1.3.1 is released, why not use the newest one ?
And for the legacy dojo 0.4.3, how shall we handle it ? Like tomcat, maitaine a our own repo ?Ideally I think we would migrate our code to up-to-date dojo. Unfortunately I have no idea how hard that would be. Does anyone? If we can't, I think there is some release of some 0.4.3 dojo, perhaps we can investigate using or repackaging it.There's also dwr.... but I think working on one dependency at a time will be less confusing.thanksdavid jencks2009/7/1 David Jencks <firstname.lastname@example.org>In my attempt to remove our svn repo I found that dojo releases a dojo-war that looks pretty similar to our repacked dojo war. I can make the build work with the substitution but I don't know enough about dojo to know if/what it breaks. Is there anyone who understands our use of dojo well enough to take a look and see if this replacement is plausible?
I recall some discussion in the distant past about not including all of dojo... I'm not sure if this is still a concern, but if the released dojo-war works and is too big we can use maven to come up with a smaller war.
See https://issues.apache.org/jira/browse/GERONIMO-4723 for my patch.