D'oh!

Sorry, this went to the old list.

-Steve



----- Forwarded Message ----
> From: Steve Schveighoffer <schveiguy at yahoo.com>
> To: Phobos <phobos at erdani.com>
> Sent: Tue, March 9, 2010 9:30:16 AM
> Subject: cast(bool) vs. opUnary!"!"
> 
> In the latest dmd 2.041, cast!T was added, and if statements were changed to do:
> 
> if(s) => if(cast!bool(s))
> 
> What happened to if translating to:
> 
> if(!!s)
> 
> which translates to ! opUnary!("!).  Hm... that's a lot of !'s :)
> 
> I thought that was the plan, at least it was in the book.  Is that going away?
> 
> It does make more sense to use cast(bool) because opUnary!"!" could potentially return a non-bool value.
> 
> -Steve

Walter and I agreed that cast to bool is as good as opUnary. Overloading unary "!" is now not possible.

Andrei

Steve Schveighoffer wrote:
> D'oh!
> 
> Sorry, this went to the old list.
> 
> -Steve
> 
> 
> 
> ----- Forwarded Message ----
>> From: Steve Schveighoffer <schveiguy at yahoo.com>
>> To: Phobos <phobos at erdani.com>
>> Sent: Tue, March 9, 2010 9:30:16 AM
>> Subject: cast(bool) vs. opUnary!"!"
>>
>> In the latest dmd 2.041, cast!T was added, and if statements were changed to do:
>>
>> if(s) => if(cast!bool(s))
>>
>> What happened to if translating to:
>>
>> if(!!s)
>>
>> which translates to ! opUnary!("!).  Hm... that's a lot of !'s :)
>>
>> I thought that was the plan, at least it was in the book.  Is that going away?
>>
>> It does make more sense to use cast(bool) because opUnary!"!" could potentially return a non-bool value.
>>
>> -Steve
> 
> 
> 
> 
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos

The only thing it disallows is making !x return something other than a bool.  The only place I can think of that would be useful is some sort of logic type that has more than 2 states, but that's a stretch.

The current implementation looks good to me, I was just curious about it.

-Steve



----- Original Message ----
> From: Andrei Alexandrescu <andrei at erdani.com>
> 
> Walter and I agreed that cast to bool is as good as opUnary. Overloading unary "!" is now not possible.
> 
> Andrei

Steve Schveighoffer wrote:
> The only thing it disallows is making !x return something other than a bool.

That's a bug. We must NOT have several redundant and potentially contradictory ways of defining interlinked features. The "!" operator is NOT overloadable, and !x must be rewritten to !x.opCast!bool().



Andrei