| re: return values from events, eventArgs
ok, I think I cleared some things up myself but I still have a number of
questions. Please correct me if I am wrong:
1) EventArgs is an object so, everytime its passed around it retains all its
data because its an object, not a int or something.
2) In order to check the progress/results from a chain of delgates (from
inheritence) is to use the GetInvocationList and then manually call them one
at a time and check the return code / EventArgs. This would happen in the
OnXXXEvent method
3) Haven't proven to myself yet but if I wanted the descendent class to
completely override the ancestor event then I need to override the
OnXXXEvent method and somehow ignore the ancestor callbacks? Not sure how
to do this, or is there an easier way? Can I ask a Delete if it is callback
to the current object or an ancestor of the current object? Makes sense in
my mind...
Questions:
A) Is it better to do all the processing based on the EventArgs or use a
return value? Is there a standard or preferred approach?
B) Should I be exposing (ie Public) the Virtual On method so that it can be
called by a button or right mouse click? or should I build an intermediate
layer or something?
C) How to do handle issue 3 above.
thx so much
"Jack Addington" <jaddington@shaw.ca> wrote in message
news:%23rT8uZMnEHA.2764@TK2MSFTNGP10.phx.gbl...[color=blue]
>I have a 3rd party object that fires an itemchanged event when someone
>edits some data on a form. This event has a custom eventArgs that has a
>field called ActionCode. In the code of the event, if you set this
>ActionCode to various values the control will respond in various ways
>(reject, reject change field, accept, etc). I am trying to understand
>exactly how that works as I am trying to extend the control and add more
>processing s events. In my previous exposure to languages, the events
>simply had return codes and I could intercept them in a descendent or
>ancestor and proccess them. Obviously it was hiding the whole idea of
>delegates and such.
>
> I could implement all my code as methods but I would like to be able to
> put the code on a form, and still be able to extend or override the
> framework code.. I've got a few problems understanding the exact
> behaviour of the EventArgs (how to return values) and also how to 'post'
> and event to fire when there is no more processing to do. Also how does
> the eventArgs really work when there are multiple subscribers? If the
> first one in the delegate thread fires should the second one be getting
> any updates?
>
> This is what I would like to do for letting the user add/insert a new row.
> There is a method on the control called InsertRow (int beforeRow){}, 0 =
> add to end.
>
> I'm going to define the AddRowEventArgs as (int beforeRow, int actionCode,
> int NewRow).
>
> 1) User hits button or right mouse to 'Add a new Row'. This calls the
> InsertRow() event
>
> event AddRow(obj Sender, AddRowEventArgs e){
> e.beforeRow = 0; (this should be set by the method calling the event
> but just in case)
> trigger event InsertRow(0)
> }
>
> event InsertRow(obj Sender, AddRowEventArgs e){
> trigger event PreInsertRow(Sender, e)
> if (e.ActionCode = ActionCode.DoNotContinue) then return;
>
> e.newRow = this.InsertRow(e.beforeRow)
>
> // run this event after all other code (descendent etc has finished
> processing)
> post event PostInsertRow(Sender, e)
> }
>
> In my 'PowerBuilder' days, I would simply have created the events with a
> return code and checked that. I know the compiler lets me do that but it
> didn't seem to clear how to capture that in .Net. Not that I'm all that
> sure how to capture the EventArgs either. Do I need to be passing them as
> ref? What I have so far is obviously a work in progress. I started using
> return values but suddenly I had concerns about what to do when 1) there
> are layers of descendants and 2) when other events subscribe.
>
> I'm starting to think is that maybe it is the OnEventFire method that is
> really suppose to be doing the logically sorting out of return code /
> eventArgs and if it finds one that returns -1 or something it jumps in an
> processes it?
>
> public delegate int DWAddRowEventHandler (Object sender,
> RowActionEventArgs args);
> public delegate int DWPreInsertRowEventHandler (Object sender,
> RowActionEventArgs args);
> public delegate int DWInsertRowEventHandler (Object sender,
> RowActionEventArgs args);
> public delegate int DWPostInsertRowEventHandler (Object sender,
> RowActionEventArgs args);
>
> public event DWAddRowEventHandler DWAddRow ;
> public event DWPreInsertRowEventHandler DWPreInsertRow ;
> public event DWInsertRowEventHandler DWInsertRow ;
> public event DWPostInsertRowEventHandler DWPostInsertRow ;
>
> protected virtual int DWPreInsertRow(Object sender, RowActionEventArgs
> args){return 0;}
> protected virtual int DWInsertRow( Object sender, RowActionEventArgs
> args)
> {
> int rc;
>
> rc = this.OnDWPreInsertRow(e);
> if (rc < 0) return rc;
>
> rc = this.InsertRow( beforeRow );
>
> this.DWPostAddRow(); //************ should this be something like
> start a new low priority thread? Seems a touch complicated to do that?
>
> return rc;
> }
>
> protected virtual int DWPostInsertRow(Object sender, RowActionEventArgs
> args){}
>
> protected virtual int OnDWAddRow( RowActionEventArgs e )
> {
> if (DWAddRow != null) {
> e.beforeRow = 0;
> return DWAddRow( this, e);
> }
> return 0;
> }
>
> protected virtual int OnDWPreInsertRow( RowActionEventArgs e )
> {
> if (DWAddRow != null) return DWAddRow( this, e);
> return 0;
> }
>
> protected virtual int OnDWInsertRow( RowActionEventArgs e )
> {
> if (DWAddRow != null) return DWAddRow( this, e);
> return 0;
> }
>
> protected virtual int OnDWPostInsertRow( RowActionEventArgs e )
> {
> if (DWAddRow != null) return DWAddRow( this, e);
> return 0;
> }
>
> Thanks so much for any help. I mostly trying to avoid do a bunch of
> useless work because I don't quite understand a couple of concepts.
>
> jack
>[/color]
| 
