www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.learn - opDispatch() is pretty damn great

reply "Szymon Gatner" <noemail gmail.com> writes:
Still exploring what D has to offer but this blew my mind:

import std.stdio;

struct Base
{
   void print(string text)
   {
     writeln("Base : " ~ text);
   }
   int add(int a, int b)
   {
     return a + b;
   }
}

struct Wrap
{
   auto opDispatch(string op, Args...)(Args args)
   {
     enum name = op;
     return __traits(getMember, base, name)(args);
   }
   Base base;
}

int main(string[] argv)
{
   Wrap wrap;
   wrap.print("wrapped call, magic!");
   auto res = wrap.add(1, 5);
   return 0;
}

I don't quite understand why "enum name" part is necessary (as my 
understanding is that "op" is compile-time constant anyway) but 
since I am crap at D that is what worked for me. I was thinking 
for log time how great something like that would be in C++ and I 
just tried this id D... Mind blown... std::reference_wrapper<> 
would be a zillion times more usable with equivalent of 
opDispatch() this powerful.
Jan 08 2014
next sibling parent reply "Adam D. Ruppe" <destructionator gmail.com> writes:
On Wednesday, 8 January 2014 at 23:20:06 UTC, Szymon Gatner wrote:

 I don't quite understand why "enum name" part is necessary


It isn't!
Jan 08 2014
parent "Szymon Gatner" <noemail gmail.com> writes:
On Wednesday, 8 January 2014 at 23:26:26 UTC, Adam D. Ruppe wrote:

 On Wednesday, 8 January 2014 at 23:20:06 UTC, Szymon Gatner 
 wrote:

 I don't quite understand why "enum name" part is necessary


 It isn't!


Indeed, I don't know why it didn't compile for me the first time.
Jan 08 2014
prev sibling next sibling parent reply "Brad Anderson" <eco gnuk.net> writes:
On Wednesday, 8 January 2014 at 23:20:06 UTC, Szymon Gatner wrote:

 Still exploring what D has to offer but this blew my mind:

 <snip>

 I don't quite understand why "enum name" part is necessary (as 
 my understanding is that "op" is compile-time constant anyway) 
 but since I am crap at D that is what worked for me. I was 
 thinking for log time how great something like that would be in 
 C++ and I just tried this id D... Mind blown... 
 std::reference_wrapper<> would be a zillion times more usable 
 with equivalent of opDispatch() this powerful.


There is also `alias this` for subtyping.

	import std.stdio;
	
	struct Base
	{
	  void print(string text)
	  {
		writeln("Base : " ~ text);
	  }
	  int add(int a, int b)
	  {
		return a + b;
	  }
	}
	
	struct Wrap
	{
	  Base base;
	  alias base this;
	}
	
	void baseOnly(Base b)
	{
		b.print("passed in a Wrap!");
	}
	
	int main(string[] argv)
	{
	  Wrap wrap;
	  wrap.print("wrapped call, magic!");
	  baseOnly(wrap);
	  auto res = wrap.add(1, 5);
	  return 0;
	}
Jan 08 2014
parent "Szymon Gatner" <noemail gmail.com> writes:
On Wednesday, 8 January 2014 at 23:27:37 UTC, Brad Anderson wrote:

 On Wednesday, 8 January 2014 at 23:20:06 UTC, Szymon Gatner 
 wrote:

 Still exploring what D has to offer but this blew my mind:

 <snip>

 I don't quite understand why "enum name" part is necessary (as 
 my understanding is that "op" is compile-time constant anyway) 
 but since I am crap at D that is what worked for me. I was 
 thinking for log time how great something like that would be 
 in C++ and I just tried this id D... Mind blown... 
 std::reference_wrapper<> would be a zillion times more usable 
 with equivalent of opDispatch() this powerful.


 There is also `alias this` for subtyping.

 	import std.stdio;
 	
 	struct Base
 	{
 	  void print(string text)
 	  {
 		writeln("Base : " ~ text);
 	  }
 	  int add(int a, int b)
 	  {
 		return a + b;
 	  }
 	}
 	
 	struct Wrap
 	{
 	  Base base;
 	  alias base this;
 	}
 	
 	void baseOnly(Base b)
 	{
 		b.print("passed in a Wrap!");
 	}
 	
 	int main(string[] argv)
 	{
 	  Wrap wrap;
 	  wrap.print("wrapped call, magic!");
 	  baseOnly(wrap);
 	  auto res = wrap.add(1, 5);
 	  return 0;
 	}


I was actually thinking about subtyping too but I don't know how 
to do alias this with opDispatch() call forwarding to Base from 
Wrap
Jan 08 2014
prev sibling parent reply "bearophile" <bearophileHUGS lycos.com> writes:
Szymon Gatner:

Two small improvements in your code:


     writeln("Base : " ~ text);


=>

      writeln("Base : ", text);



 int main(string[] argv)
 {
 ...
   return 0;
 }


=>


 void main(string[] argv)
 {
 ...
 }


Or even:


 void main()
 {
 ...
 }


Bye,
bearophile
Jan 08 2014
parent reply "Szymon Gatner" <noemail gmail.com> writes:
On Wednesday, 8 January 2014 at 23:46:28 UTC, bearophile wrote:

 Szymon Gatner:

 Two small improvements in your code:


    writeln("Base : " ~ text);


 =>

      writeln("Base : ", text);



 int main(string[] argv)
 {
 ...
  return 0;
 }


 =>


 void main(string[] argv)
 {
 ...
 }


 Or even:


 void main()
 {
 ...
 }


 Bye,
 bearophile


Thanks, was not aware, main() was actually generated by VisualD
Jan 08 2014
parent "Jakob Ovrum" <jakobovrum gmail.com> writes:
On Thursday, 9 January 2014 at 00:02:00 UTC, Szymon Gatner wrote:

 On Wednesday, 8 January 2014 at 23:46:28 UTC, bearophile wrote:

 Or even:


 void main()
 {
 ...
 }


 Bye,
 bearophile


 Thanks, was not aware, main() was actually generated by VisualD


Yeah, VisualD was guilty of propagating that poor template for a 
long time. It's now fixed.

http://d.puremagic.com/issues/show_bug.cgi?id=11277
Jan 08 2014