Re: [Pnet-developers] Should this code compile or now ?

[Gopal V]

If memory serves me right, Rhys Weatherley wrote:
> > Good luck trying to fix this - you'll probably have to make the operators
> > modal with an ILGenInfo flag for "evaluating an enum constant".  Ick!
> 
> Actually, an easier way to fix this may be in the lvalue code that fetches 
> constants (the call to FieldToConstant in SemToLRValue).  If the compiler is 
> in the "evaluating an enumerated constant value" mode, then return the 
> underlying type in the CSSemValue, not the enumerated type (this is around 
> line 578 of cs_lvalue.tc).
> 
> That will probably be better than hacking every operator to recognise 
> enumerated values as integers.

Hmm... this would make more sense . But this would mean every compiler
dev has to build it himself :-)

I think I fixed it with a couple of lines and no ILGenInfo option :-)

And if I forgot to mention , Dinosaurs d=Dinosaurs.Rex + Dinosaurs.Stegy 
also fails to compile for pnet ... (but since 99% of all enum operations are
either '|' or '&' , this is a rare case ;-)

Going on to fix that ... (and an initializer overflow bug , like 
Int32.MaxValue+1 for an int32 enum initializer compiles as of now).

Gopal
-- 
The difference between insanity and genius is measured by success