I am from china and poor of english,i am sorry you may can't
understand what i am meaning.
Recently,i am study a lite about haskell,so i think now D's
template is some like a functional program evolution form c++.
because i find some restrict,i think it's still in a low-level
for we still think it in a c++ way.Should we drop more from c++,
then go into a high-level template program?
now i want to point out what i found which is can't tolerance and some
i think should be work out.
1:
this code don't work because the name query somelike not a bidirection
way,[alias ReturnType!(T) Out],can't resolve the opCall
class monad(alias T)
{
alias ReturnType!(T) Out;//*****
}
class maybe(T):monad!(maybe!T)
{
auto opCall(maybe!T)
{
alias maybe!long R;
return R.init;
}
}
there is another sample:
struct B(T)
{
}
alias B!(C) BC;//identifier 'C' is not defined!
alias B!int C;
2:
add parital template feature
struct P(A...)
{
}
alias P!int P1;
P1 p1;//instance of P!int
alias P1!long P2;
P2 p2;//intance of P!(int,long)
this feature will make D has the power to write a template functional
library,and change D into a high-levle program language.
Comment #1 by bearophile_hugs — 2010-10-25T15:51:38Z
(In reply to comment #0)
> I am from china and poor of english,i am sorry you may can't
> understand what i am meaning.
I think we are able to understand what you mean.
> this code don't work because the name query somelike not a bidirection
> way,[alias ReturnType!(T) Out],can't resolve the opCall
>
> class monad(alias T)
> {
> alias ReturnType!(T) Out;//*****
>
> }
>
> class maybe(T):monad!(maybe!T)
> {
> auto opCall(maybe!T)
> {
> alias maybe!long R;
> return R.init;
> }
> }
This seems a valid bug report or enhancement request.
> there is another sample:
> struct B(T)
> {
> }
> alias B!(C) BC;//identifier 'C' is not defined!
> alias B!int C;
This is a bug report, and I think it's already present somewhere in Bugzilla.
In future I suggest you to put just one bug report (or enhancement request) for each Bugzilla entry.
> 2:
> add parital template feature
> struct P(A...)
> {
> }
> alias P!int P1;
> P1 p1;//instance of P!int
> alias P1!long P2;
> P2 p2;//intance of P!(int,long)
>
> this feature will make D has the power to write a template functional
> library,and change D into a high-levle program language.
This is another different enhancement request. I have never seen it before. It looks cute. Are you able to show one or more possible usages of it?
I will soon link this bug report in the main D newsgroup.
Comment #2 by galaxylang — 2010-11-04T22:13:55Z
I have not on net for a long time,because i have not a network at home,
(network is still very expensive in our country in our country :)
and i am now prepare for some examine.so i submit a new issue to attract
your notice,and some new i advise.
http://d.puremagic.com/issues/show_bug.cgi?id=5067
Comment #3 by christoph — 2014-01-09T03:31:58Z
D is actually able to do something like partial templates.
Here is an example that works:
struct P(A...)
{
alias Args = A;
}
alias P!int P1;
P1 p1; // P!int
alias P!(P1.Args,long) P2;
P2 p2; // P!(int, long)
void main() {
writeln(typeid(P2)); // Output: main.P!(int, long).P
}