Not sure about a) outside of using reflection and Invoke(), but you can
retrieve all necessary data about types and their methods, etc using a
combination of Type.GetType, Assembly.GetType, and System.Reflection.
This is used extensively in both mcs and MonoLOGO.

I take it you're wrapping ORBit and I'll be able to wrap libbonoboui in
Gtk# soon? :-)

-- Rachel

On Wed, 2002-06-26 at 10:58, Michael Meeks wrote:
> Hi there,
> 
>       Apologies for arriving so late with my clueless questions; anyway -
> I've been poking around with delegates, and they seem quite pleasant,
> but inadequate for what I need.
> 
>       I suppose I need to know how to do 2 things, and search as I do I can't
> seem to see how to do them:
> 
>       a) A way to specify that a certain object is extremely late 
>          linked, such that I can invoke arbitrary methods on it, and
>          the compiler will do ~no checking eg.
> 
>       LateLinkType t = (LateLinkType) myObject;
> 
>       t.unknownMethod ("astring");
> 
>          The above being 'useful' for a stubless ORB. Of course, 
>       perhaps it's possible that I have to have all the type 
>       information there at compile time and it's not possible to have
>       a stub-less CORBA wrapper. I can see how the skel-less process
>       works, we do a GetType (), find the MethodInfo, and do an Invoke
>       but how should the stub case work ?
> 
>       b) A way to handle delegates that is elegant; eg. I wish to
>          implement the following (or similar):
> 
>       private void sizeAllocate (Widget w, Allocation a,
>                                  Object closure);     
>       ...
>               w.addHandler ("size_allocate", sizeAllocate, myObject);
> 
>          I don't mind doing a (redundant) new Foo (sizeAllocate), and 
>       I can see how you can do this with loads of delegates: new 
>       DelegateWidgetAllocation (sizeAllocate) eg. but I don't want to
>       go around typing the method signature twice - once in declaring 
>       it, and once in a delegate allocation scheme.
> 
>          Is it possible to coerce the type information out of a method
>       name by some clever casting ? and if so, how in the (seemingly 
>       unexpandable) delegate constructon scheme ?
> 
>       Seemingly I've missed the plot somewhere here :-)
> 
>       Thanks in advance,
> 
>               Michael.
> 
> -- 
>  [EMAIL PROTECTED]  <><, Pseudo Engineer, itinerant idiot
> 
> 
> _______________________________________________
> Mono-list maillist  -  [EMAIL PROTECTED]
> http://lists.ximian.com/mailman/listinfo/mono-list
> 




_______________________________________________
Mono-list maillist  -  [EMAIL PROTECTED]
http://lists.ximian.com/mailman/listinfo/mono-list

Reply via email to