Can someone explain why I can't instantiate this template implicitly andor comment on whether I'm overlooking a better way to accomplish what this template does?

template increment(T, K)
{
void increment(T[K] association, K key)
{
if(key in association)
{
association[key]++;
}
else
{
association[key] = 1;
}
}
}

Thanks,
Anthony

anthony.difranco@yale.edu wrote:
> Can someone explain why I can't instantiate this template implicitly andor
> comment on whether I'm overlooking a better way to accomplish what this template
> does?

Mostly because ITI is still incomplete.  Anything beyond strict type replacement doesn't really work.


Sean

In article <dvvk0i$2tum$1@digitaldaemon.com>, anthony.difranco@yale.edu says...
>
>Can someone explain why I can't instantiate this template implicitly andor comment on whether I'm overlooking a better way to accomplish what this template does?
>
>template increment(T, K)
>{
>void increment(T[K] association, K key)
>{
>if(key in association)
>{
>association[key]++;
>}
>else
>{
>association[key] = 1;
>}
>}
>}
>
>Thanks,
>Anthony
>
>

IIRC

if(T* val = (key in association))
(*val)++;
else
association[key]=1;

haven't tried it but it should work

BCS wrote:
> In article <dvvk0i$2tum$1@digitaldaemon.com>, anthony.difranco@yale.edu says...
>> Can someone explain why I can't instantiate this template implicitly andor
>> comment on whether I'm overlooking a better way to accomplish what this template
>> does?
>>
>> template increment(T, K)
>> {
>> void increment(T[K] association, K key)
>> {
>> if(key in association)
>> {
>> association[key]++;
>> }
>> else
>> {
>> association[key] = 1;
>> }
>> }
>> }
>>
>> Thanks,
>> Anthony
>>
>>
> 
> IIRC
> 
> if(T* val = (key in association))
> (*val)++;
> else
> association[key]=1;
> 
> haven't tried it but it should work

Yes. A full working version with the current limited IFTI would look like:

template increment(T,K) {
	static assert(is(K : typeof(T.init.keys[0])));
	void increment(inout T association, K key) {
		if(auto val = (key in association))
			(*val)++;
		else
			association[key]=1;
	}
}


/Oskar

Thanks; it works.

In article <e00fq4$131r$1@digitaldaemon.com>, Oskar Linde says...
>
>BCS wrote:
>> In article <dvvk0i$2tum$1@digitaldaemon.com>, anthony.difranco@yale.edu says...
>>> Can someone explain why I can't instantiate this template implicitly andor comment on whether I'm overlooking a better way to accomplish what this template does?
>>>
>>> template increment(T, K)
>>> {
>>> void increment(T[K] association, K key)
>>> {
>>> if(key in association)
>>> {
>>> association[key]++;
>>> }
>>> else
>>> {
>>> association[key] = 1;
>>> }
>>> }
>>> }
>>>
>>> Thanks,
>>> Anthony
>>>
>>>
>> 
>> IIRC
>> 
>> if(T* val = (key in association))
>> (*val)++;
>> else
>> association[key]=1;
>> 
>> haven't tried it but it should work
>
>Yes. A full working version with the current limited IFTI would look like:
>
>template increment(T,K) {
>	static assert(is(K : typeof(T.init.keys[0])));
>	void increment(inout T association, K key) {
>		if(auto val = (key in association))
>			(*val)++;
>		else
>			association[key]=1;
>	}
>}
>
>
>/Oskar