[Mono-list] removing delegate

[Daniel Stodden]

On Sun, 2002-08-04 at 01:44, YUAN-FEN KUO wrote:

[cc'd to mono-dev]

> I think the follwoing piece of code will make my question clear:
> 
> 
> delegate void delA();
> delegate void delB();
> 
> class main{
>         static void Main(){
>                 delA food;
>                 delA aoo = new delA(aa);
>                 delB boo = new delB(aa);
>                 delA coo = new delA(aa);
> 
>                 //legal combine
>                 food += aoo;
>                 food();
> 
>                 //combine wrong type
>                 //food += boo;   //throws exception as expected during runtime, though I 
> wonder why not compile time
>                 //food();
> 
>                 //revomve wrong type
>                 food -= boo;   //does not throw any exceptions though type 
> mismathched??
>                 food();
> 
>                 //legal remove 
>                 //"Attempting to remove a delegate from an empty list (or to remove a 
> non-existent delegate from a non-empty list) is not an error" is said in the last 
> paragraph of Secion 15.3
>                  //"Impossible Removal is benign" was used in their sample code in 
> Sec. 15.3 as well.
>                 food -= coo;
>                 food();
> 
>         }
> 
>         static void aa(){System.Console.WriteLine("method aa");}
> 
> }

just tried it. this is indeed helpful.

first, there is a bug in mcs. removing the wrong type (food-=boo) should
not be allowed by the compiler.

second, no mscli does not throw on food-=boo. it never does as i pointed
out. but what's _really_ interesting here is that (given an actual
assembly as written above) food -= boo actually suceeds, leaving you
with a null ref.

running mono, the second food() gets you a second call to aa().
running mscli, the second food() already bails out on null.

i'll have to look into this. you may file a bug for the mcs gang in the
mean time if you want to [if there isn't one already].

thanks a lot, i hate you ;),
daniel

> > > It seems that type mismatching isn't caught when removing a 
> > delegate from a 
> > > sequence of delegates.  
> > > I am just wondering if this were done to match the Spec's saying 
> > that 
> > > "impossible removal is benign."

-- 
___________________________________________________________________________
 mailto:stodden@in.tum.de