[Design of JBoss jBPM] - Re: splitting the work on the console
by kukeltje
There are several things that should/could be done. One is to decide if there should be some backwards compatibility (functionality) in the 3.2 webapp. Especially the generated form. IMO it should at least have an automagically generated form. For the simple reason that it gives me a real quick fully functional process to test. I do not want to focus om the forms in the beginning. (see my commments on this in the different jira issues)
The other thing is to port the admin, or rather the monitor page to facelets.
Third thing is that there should be a simple transition button tag (or maybe simplify the way it works)
Forth is that the forms should be versionable (currently if you have two forms in two different processes (or even versions of processes) they clash, or rather, the one that was first wins.
Funny thing is, that this is a clean separation of 'task' and 'monitor' menu items.
Oh, yes, some obsolete/orhaned code should be removed from the webapp.
View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=3965083#3965083
Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=3965083
19 years, 7 months
[Design of JBoss jBPM] - splitting the work on the console
by tom.baeyens@jboss.com
"kukeltje" wrote : | I was just wondering about the progress? There are some
| | issues assigned to me would probably be the first things you,
| | David, could be working on. Since I have no idea what is
| | going on, and I have other things to do, I did not start
| | working on the webapp based issues.
|
there is no internal communication outside of this forum about development progress.
Use this dev forum as your main mechanism to coordinate work. If afterwards it turns out that jira is useful, we could switch to that. But for now, start with discussing here.
Another thing we could use to organise work is by working on different menu items. But in the backing beans, we should get some merges in case of concurrent modifications. Should be possible to work like that, i think.
what do you guys think
View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=3965076#3965076
Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=3965076
19 years, 7 months
[Design of JBoss jBPM] - Re: Anyone created a RMI or Webservice call wrapper?
by kukeltje
I'm working on one or at least will be on short notice. Heck, I even might have complete running functionality before the end of the month.
My ?0.02:
Do NOT use the SLSB, no need (imo) and it makes things NOT work on Tomcat. There is a company that wants a webservice frontend, but they just run Tomcat. Instead of using the/a SLSB, the commandexecutor can be used (not the thread, just the class). I just tried it and it works.
The mapping of the commands (creating an xsd for this) is a jira issue. The commands can be realy simple Most calls can even be just rpc type calls, very easy to generate.
Starting a process (just by name or with a version) was implemented in 1 minute (hour is more like it ;-) ) after I got the basics running. Other plain rpc type calls for ending a task (with or without variables, with or without a transitionname) is also not difficult. The thing is that there is no method overloading in wsdl (at least will not be supported in future versions afaik), so for starting a process I have two methods, startProcessLatestVersion (..) and startProcessSpecificVersion(..). Notr beautiful, since with the end task there will be numerous combinations (see also the jira, starting from http://jira.jboss.com/jira/browse/JBPM-718)
More complex things will be retrieving a tasklist, logging etc...
My preference is that someone defines a nice XSD and we map that to the command objects, extending those where needed. I'm not a fan of generating xsd's from java (although it did work with JBossWS generated wsdl/webservice and an the eclipse webservice explorer)
View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=3965069#3965069
Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=3965069
19 years, 7 months
[Design the new POJO MicroContainer] - Re: Bean instantiate order with contextual injection
by adrian@jboss.org
"alesj" wrote : anonymous wrote :
| | 1) Contextual injection - i.e. determining (and properly dependening on)
| | the injected context based on class rather than name.
| |
|
| How to get property class to add 'demand' dependency item in AbstractInjectionMD - AbstractPropertyMD doesn't hold any class type value - obviously, since no owner bean class inspection is done?
|
Ok. Your problem is that the class is not available until the
DescribeAction. That is when the classloader dependency is
resolved. Until that point we can't guarantee the classloader even
exists.
So you're going to need a "placeholder" DependencyItem
where you fill in the type of the property once the BeanInfo
is resolved in DescribeAction.
This is a bit messy, and would probably be better implemented
as a "revisit" of the visitor where the extra visit does extra work
once the BeanInfo is resolved, but that is a lot more work
(in terms of making the old understand it rather than doing work twice).
For now, I would just do context.getDependencyInfo(),
iterate and update the "placeholder"s.
anonymous wrote :
| anonymous wrote :
| | 2) Creating BeanMetaData from annotations rather than xml.
| |
|
| Adding this in AbstractBeanMetaData.visit() through KernelConfigurer?
See above. And see where annotations and their dependencies
from AOP are resolved in the DescribeAction.
View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=3965028#3965028
Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=3965028
19 years, 7 months