Re: [ve-dev] RE: Need Help

[Dr Gili Mendel <gmendel@xxxxxxxxxx>]

In general every feature will have a
decoder and _expression_ associated with it.  The question is not just
on the top/down generation side, but what should be done bottom up... i.e.,
if you have no code associated with
the feature, will you need to reCreate it  when you parse the code
in and it that case how?

An example of this will be in the SWT
case, where we have two features:   controls, allocation.

allocation, the feature that denotes
how to instantiate the object and is related to the Constructor _expression_
for all JavaBeans

        MyControl
= new MyControl(myShell, SWT.NONE);

The controls feature is the SWT feature
that designate the container/child relationship (e.g., a button inside
a composite), and will determine, within others, the z order of the child.

In some cases, if the Control is a composite
we create an init method for the Control, so the control feature will be
associated with the following _expression_:

private
void
createMyShell() {
       myShell
= new Shell();        
         createMyControl()
<------------------------  controls _expression_ will determine the
z order
       

}


private
void
createMyControl() {
        .....
           
    MyControl = new MyControl(myShell, SWT.NONE);    
                   <--------------allocation
_expression_ ... constructor feature and will be used by the target VM

}


In other cases, when the Control is
not a composite, we will reUse the container's init method:


private
void
createMyShell() {
       myShell
= new Shell();        
         MyControl
= new MyControl(myShell, SWT.NONE);          
             <--------------allocation
_expression_
       

}


.... here we get a case where there
is NO code associated with the controls feature, but we need to create
it during parsing.... in the later case the allocation _expression_ drives
the z order on behalf of the controls feature.


CodeGen does create a CodeExpressionRef
for the control feature in the last case, but it marks it with the
STATE_NO_SRC
flag.

... this area is changing with https://bugs.eclipse.org/bugs/show_bug.cgi?id=72561
... but check out the CompositeAddDecoderHelper.





The isChildValue() is being used to
determine if the feature refers to a value that potentially have a instance
variable or a local variable associated with it.






------------
Dr. Gili Mendel
IBM
Software Development
RTP Raleigh, NC
(919)543 6408, tie: 441 6408

"Janak Mulani" <janak.mulani@xxxxxxxxx>
Sent by: ve-dev-admin@xxxxxxxxxxx

09/01/2004 04:46 PM

Please respond to
ve-dev

To	Dr Gili Mendel/Raleigh/IBM@IBMUS
cc	"Sibylle. Peter@Canoo. Com" <sibylle.peter@xxxxxxxxx>, "Joe Winchester" <WINCHEST@xxxxxxxxxx>, Srimanth Gunturi/Raleigh/IBM@IBMUS, Rich Kulp/Raleigh/IBM@IBMUS, <ve-dev@xxxxxxxxxxx>
Subject	[ve-dev] RE: Need Help

Dear VE Team:

We have a Structural feature setting on a bean for which No Source should
be
generated.

Please refer to the "containment" SF in the attached override
file.

1. How should we handle this?

2. In the following method of ObjectDecoder, for the above SF we get
isChildValue = true. What does this mean?

protected boolean isChildValue(EStructuralFeature sf, IJavaInstance val,
boolean checkAttributes) {
    if (sf == null || val == null || !(val instanceof IJavaObjectInstance))
return false ;
    // Known SF to be ignored
    if (MethodTextGenerator.isNotToBeCodedAttribute(sf,(EObject)val))
return
false ;
    // Known Instances to be ignored
    if (isSimpleObject((IJavaObjectInstance)val)) return false
;
    if (checkAttributes)
      if (! isAnyAttributeSet((IJavaObjectInstance)val))
return false ;

    return true ;

3. For the above SF we would like the
MethodTextGenerator.isNotToBeCodedAttribute(sf,(EObject)val) to return
true
because we donot want any source generated while setting a value on the
SF.

I hope I have provided sufficient info for you to answer the questions.

Thanks and regards,

Janak

Attachment: ULCComponent.override
Description: Binary data