First off, I would like to thank everyone for their quick reply to Justin's
original post. J My name is Ian Bieder and I also work for Nexteppe. I am
also a Flex newbie plus I just graduated from GaTech a few months ago. I am
really excited about Developing in Flex and from what I have seen so far
Flex looks very promising. Based on the many responses to Justin's original
message, he has asked me to elaborate a little more on how we have things
setup and our problem.
We are using WebOrb to make remote calls to our .NET backend for our app.
In Flex we are using an MVC design. We have a singleton controller object
that initiates most of the major UI changes, remote data calls, and holds
references to most of the model objects (note: this singleton controller
will be broken up as our app gets larger). For example, we have a class
called "UserProvider" that handles some remote object calls and it is
declared in the controller. We initiate remote methods by calling things
like aUserProvider.getAUser(userID). Once aUserProvider receives data back
it stores it in variable that is method specific and for example we can
access it in the controller by calling aUserProvider.result. aUserProvider
also will fire off and event once it gets a result set back from the remote
service so that the controller knows that new data has been returned.
Our problem arises when we would like to make use of methods in UserProvider
for multiple things in the UI. The way we have things setup now gets the
job done but I am really unhappy with it and I foresee it causing us
problems in the future. For example, we have a part of the UI where we would
like to view a quick summary profile of a user and another where we would
like to edit the profile of a user. For both UI operations we need to pull
all of the information about a single user (i.e. call
aUserProvider.getAUser(userID)). This is sort of how we have the code setup
now in the controller (I renamed, omitted some stuff like fault handling
etc., and made some small changes to make my code more understandable for
this example):
Once the "view user profile summary" button is clicked this method gets
called in the controller:
public function beginSwitchToUserSummaryView(): void
{
summaryViewWaiting = true;
userProvider.getAUser(("user to get complete profile
from").userID);
}
Once the "edit user profile" button is clicked this method gets called in
the controller:
public function beginSwitchToUserEditView(): void {
editViewWaiting = true;
userProvider.getAUser(("user to get complete profile
from").userID);
}
Once a result set is returned from userProvider after calling "getAUser()"
an event is fired that the controller catches and then the controller calls
this method that starts the necessary UI changes:
public function aUserWasReturned(): void
{
if (summaryViewWaiting) {
switchToUserSummaryView(userProvider.result);
//tells the UI to now display the user in the summary view
summaryViewWaiting = false;
}
if (editViewWaiting) {
switchToUserEditView(userProvider.result);
//tells the UI to show the user in the edit view
editViewWaiting = false;
}
}
Is there an easier way to reuse the same remote object call (i.e.
getAUser()) for multiple UI operations? I hate using flags like this and I
am fairly sure it is going to cause problems as the app grows. Also, is
there some sort of other way that we should be trying to implement a MVC
approach (Yes, I know, very generic question J)? Again, thanks for
everyone's helpful suggestions so far.
UPDATE: Based on the responses so far it looks like I should be passing some
information with my remote call that describes what method is doing the
calling. That way I can pass that data back and handle the returned data
accordingly. Also, Scott Talsma suggested that I take a look at Cairngorm,
because it handles a lot of this ,and I will definitely be doing so.
Thanks,
--
Ian
Nexteppe Automotive Automation
(877) 572-8892 support
www.nexteppe.com
---------- Forwarded message ----------
From: Justin Nichols <[EMAIL PROTECTED]>
Date: May 24, 2007 10:19 AM
Subject: [AFFUG Discuss] MVC, Object Instantiation/Population, Asynchronous
Calls
To: [email protected]
First off I'd like to introduce myself. My name is Justin Nichols and I
work for Nexteppe Business Solutions and we deal in car dealer data/web
services. We're working on a new system using Flex, WebORB, and a .NET
backend. We're running into a conceptual issue concerning the asynchronous
calls being made when retrieving remote data. We are new to the Flex game
so please forgive any ignorance I may show :)
Here's the basic problem and I have a feeling about what the solution would
be, but I wanted to ask:
1) We are setting up our system using MVC as best we can
2) We have written some .NET backend objects, and for this example let's say
we're needing to populate an ArrayCollection (flex-side) of User objects.
3) We have our controller set to retrieve, using a RemoteObject call, an
ArrayCollection of User objects called userList.
4) The scenario we're in is the following:
a) We have a button on the screen that when clicked we want to get the
userList and then perform some action using the userList.
b) The button onclick event will send the necessary call to the
controller which will in turn make the call to the server and retrieve the
list.
c) Immediately after the call is made to populate the userList variable,
another statement is written that will perform some action on the userList
(say userList.getItemAt() or whatnot).
d) The issue at hand, is that because the call is asynchronous, the
userList may very well be null when the userList.getItemAt() gets called.
Now from what I've read, we simply need to set up a custom event that gets
dispatched once the userList object has been filled. However, the problem
then arises when we might have two separate UI operations that would utilize
the same userList object population methodology, so my main question is,
would we need to write an event handler for the population of the userList
object for each UI operation (thus knowing what operation I'm really wanting
to do once the userList object is filled)? Or is there some other best
practice technique that I'm missing here?
Also, is this an issue that frameworks such as Cairngorm can help with?
I really appreciate all the help in advance and again I hope I'm not being
too...noobish :)
Thanks,
--
Justin Nichols
Nexteppe Business Solutions
(877) 572-8892
[EMAIL PROTECTED]
-------------------------------------------------------------
To unsubscribe from this list, simply email the list with unsubscribe in the
subject line
For more info, see http://www.affug.com
Archive @ http://www.mail-archive.com/discussion%40affug.com/
List hosted by http://www.fusionlink.com
-------------------------------------------------------------
