On Thu, 18 Jan 2001 Christopher Watson wrote:
> Is there a better way to implement a class variable in a parent script?
>
> And, is the answer going to be "use a global"? I hope not, because I don't
> want to.
>From what I understand from your explanation, I believe you are referring to
the ancestor property.
If every time you declare a child object, you declare the parent as the
child's ancestor, the properties of any objects higher in the hierarchy can
be accessed without specifying the whole 'path' to that particular object.
i.e.
--- parent object:
myChildObject = new (script"childObject",me)
--- child object:
property ancestor
on new me,whichAncestor
ancestor = whichAncestor
return me
end
Then, at any given child object, you can access any properties of the root
object (or any other child on the branch if a property with the same name
exits), with:
me.classVariable
Director will start looking in the same instance, but if the property is not
found, it will 'take the walk up the linked list', until it finds it.
In your case, define your 'class variables' in your root object, all its
children will be able to access them, by preceding the name of the property
by 'me'. Simply make sure to assign unique names to these variables.
I don't know if OOP purists agree with this approach, since you are
accessing an object's properties from the outside, but I've used it quite
often with success, specially for the tree-shaped structures formed when
parsing an XML document.
Let me know if this answers your question.
victor
> The properties within a parent script are equivalent to instance variables.
> But I want to implement the equivalent of a class variable in my parent
> script, so that all child objects created from that parent script can take
> into account the value held within that class variable when performing tasks
> within their instance methods.
>
> Now, in my particular case, I'm not creating discrete child objects that
> each live in a world of their own. All the child objects created from my
> parent script are nodes within a tree (an XML document), with one of those
> child objects representing the root of that tree. The root has child nodes,
> and those child nodes have child nodes, etc. Taking that into consideration,
> I though that I had come up with a fairly simple way of faking a class
> variable.
>
> If I were to maintain the tree as a linked list (meaning, if the parent
> script had a property which held a reference to the child object which is
> its parent node), I would always be able to recurse to the root from any
> node and retrieve the value held within the "class variable" property in the
> root node child object. Since the root node child object never changes for
> the existence of the tree, setting a property value in the root would be the
> same as setting the value for all the children, as long as they all went to
> the root to look for that value.
>
> The problem with this approach is that every time a node is called upon to
> execute its method for forming the textual version of its part of the XML
> document, it must take that walk up the linked list to get the value of that
> property from the root node. It's the same walk for every single element
> within an XML document. And you know how complex XML docs can get.
>
> It works. But it seems terribly ineffient to me.
[To remove yourself from this list, or to change to digest mode, go to
http://www.penworks.com/LUJ/lingo-l.cgi To post messages to the list,
email [EMAIL PROTECTED] (Problems, email [EMAIL PROTECTED])
Lingo-L is for learning and helping with programming Lingo. Thanks!]
