Igor Stasenko wrote:
> 2428 run, 2406 passes, 0 expected failures, 15 failures, 7 errors, 0
> unexpected passes
> 
> 
> After applying changes to sets using nil wrappers [1]:
> 
> 2428 run, 2406 passes, 0 expected failures, 15 failures, 7 errors, 0
> unexpected passes
> 
> 
> After adding changes to sets using negative tally[2]:
> 
> 2428 run, 2406 passes, 0 expected failures, 15 failures, 7 errors, 0
> unexpected passes

Those are great results!

> [1] http://bugs.squeak.org/file_download.php?file_id=3829&type=bug

Yeah... seeing the code I like the wrapper solution even better. It's 
just so elegant. Virtually no overhead, nicely dealing with all sorts of 
nestings, having the option for future extensions (weak elements, 
collection elements etc). I think I've just promoted that to my top 
choice ;-)

Seriously folks, look at that code. It's a great solution.

Cheers,
   - Andreas