Both is possible. However, in our practice we try to test as much as possible. The call to a cairngorm command is rather difficult to test, whereas you can easily test if a ciarngorm event is dispatched by adding a listener to the CairngormEventDispatcher.
Cheers Ralf. On Nov 16, 2007 12:34 PM, Yigit Boyar <[EMAIL PROTECTED]> wrote: > > > > > > > hi all; > i work in a project which includes the implementation of a RIA > application in which the user will generate an XML based on an XSD > defined language using drag&drops and etc. (user never sees the xml). > > at first, we developed a code generator using AIR that takes the XSD and > generates vo's, commaders, events,model & controller. so we just hand > coded the views. and the demo was very successful. > > now the second phase of the project came and there are huge differences > in the XSD, so we had to handcode additional things into commanders, > which violated the auto-generation approach and resulted in some bugs & > inefficiency. then we decided to upgrade the code generator to create > more clever commanders and more specific events. > > we have an important debate here for which we need suggestions, so i'm > presenting you our question. > > when an event is being handled, there may be need to do some more things > which is done by another commander and has a specific event. in such a > situation, > -- should the commander call the other commander just creating the event > and sending it to the other's execute method > or > --should the commander create the event and just dispatch. > > the second is the normal way for the cairngorm specs but the first is > much more efficient. > so which one do you suggest? > > thnks. > > p.s. commanders are specific to VO's a a commander just edit's it's VO's > data on the model. > -- Ralf Bokelberg <[EMAIL PROTECTED]> Flex & Flash Consultant based in Cologne/Germany