Peter Donald wrote:
>On Sun, 12 May 2002 09:51, Stephen McConnell wrote: > >>>* lifecycle "style": Is it poolable, is it re-entrant, is it threadsafe, >>>is singlethreaded etc >>> >>Just a note that I think that we will sooner or later have to deal with >>the pooled concept at the framework - by the addition of a specific >>interface for a Pool, >> > >Thats how it used to be back a while ago. We used to have Poolable and Pool >interface inside the framework proper. Thankfully that got moved out. > > >>or, elimination of pooling notions at the level of >>ServiceManager and ComponentManager. >> > >+1 > Thats my preference as well. (But we know that this is am emotive topic!). > >>>* context: >>> - Context Class >>> - Entrys in Context (both name of entry and type of entry) >>> >>I'm assuming you mean something along the lines of >>excalibur.context.ContextUtility (which I'm using within Merlin). It >>handles the above but the javadoc really needs some work. >> > >Very similar. However it needs to be more like J2EE <resource-ref/> in that we >can map arbitrary types into context. In particular I would like to be able >to map objects of type MBeanServer into context. > >I also plan to use the converter architecture (in excalibur now, previously of >myrmidon) to help with populating the context. Basically converter >architecture tries to convert from one type to another type. > Can you give me some liks/references etc. to <resource-ref/> and the converter stuff > > >>>I was going to try address it in a global manner first but if you want me >>>to make it possible for you to overide default behaviour with >>>CascadingConfigurerr then it should be relatively easy. Just say the word. >>> >>Before answering the above, I'll go into a more detail concerning Merlin >>configuration semantics. >> >...snip... > There you go again ... snipping the best bits of my emails! :-) > > >sounds interesting - I will have a further look. I am not sure thats the right >way to go with all phoenix but it should be possible to enable by just >writing a new Deployer. Wait till next weekend (after I put through all >changes I proposed) and hopefully it will be much easier to enable this sort >of thing. > No problem. >Currently it is a PITA because we are supporting a bunch of >deprecated deployment formats (and thus need similar code in multiple >places). After this is all cleaned up it should be much easier to do. > > >>For completeness, there are two other .xinfo additions that should be >>noted. Firstly, there is the introduction of the name attribute on the >>block element. >> > >+1 > > >>Secondly, Merlin supports a >><context> declarations within a .xinfo which is handled by the >>ContextUtility class. >> > >+1 to idea but I want to play with concept a bit more because I need other >things to be placed in context that wont be supported by current ContextUtil. > > Agreed. Cheers, Steve. -- Stephen J. McConnell OSM SARL digital products for a global economy mailto:[EMAIL PROTECTED] http://www.osm.net -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
