[Pnet-developers] Bug , Patch and doubts

[Gopal V]

Hi,
        This is probably the simplest stuff I must have had doubts about.
To be more precise "have doubts" as well...

So postulates (which I assumed).

*) No attribute argument is a single identifier 
*) The only way a single identifier can be semanalysed inside an attr arg
   is if it is part of a qualified identifier
*) Considering all the possible qualified identifiers in the attribute 
   args , they'll all be member accesses . So obviously , whenever 
   CSResolveSimpleName is called inside ILNode_ToAttrConst , we will need
   either a namespace or type but never a class member . So avoid them in
   CSResolveSimpleName :-)

   .. and hence the proof ..

So the attached patch & code should make things a lot more clear ...

Gopal
PS: writing under the influence of too much maths and coffee ...
-- 
The difference between insanity and genius is measured by success

patch
Description: Text document

PropertyAndAttribute.cs
Description: Text document