• Timothee Cour via Digitalmars-d-learn (28/28) Jun 21 2015 Suppose I have:
• "Marc =?UTF-8?B?U2Now7x0eiI=?= <schuetzm gmx.net> (17/55) Jun 22 2015 It's possibly not a real Voldemort type, i.e. one that closes

Timothee Cour via Digitalmars-d-learn <digitalmars-d-learn puremagic.com> writes:

Suppose I have:

import std.range;
auto a=iota(complex_expr_returning_3());

I'd like to have a function/trait/template/compiler magic that takes
variable a and generates a string that can be mixed in to represent the
type of a. The difficulty is that typeid(a).to!string doesn't work for
Voldermort types:
typeid(a).to!string produces "std.range.iota!(int, int).iota.Result", which
is not very useful as it can't be mixed in a program, being a voldemort
type, eg:
mixin("std.range.iota!(int, int).iota.Result a;"); //compile error

However, the compiler could help reduce it to something as simple as
possible, eg:
"typeof(iota(int.init))"
here this would work:

mixin("typeof(iota(int.init)) a;"); //works

Is it possible to do that in a generic way, such that it hides as much as
possible the details of the expression (here, "complex_expr_returning_3()"
should be simplified to int.init)

Or is there another trick I could use to instantiate a variable with same
type as a?


NOTE:
typeof(a) a2;
works even though
http://www.drdobbs.com/cpp/voldemort-types-in-d/232901591?pgno=2 says it
won't work, "Sorry, that won't work, the compiler will not allow a
Voldemort Type to be instantiated outside of its scope "
(although that's doesn't help with my original problem)

"Marc =?UTF-8?B?U2Now7x0eiI=?= <schuetzm gmx.net> writes:

On Sunday, 21 June 2015 at 23:25:47 UTC, Timothee Cour wrote:
 Suppose I have:

 import std.range;
 auto a=iota(complex_expr_returning_3());

 I'd like to have a function/trait/template/compiler magic that 
 takes
 variable a and generates a string that can be mixed in to 
 represent the
 type of a. The difficulty is that typeid(a).to!string doesn't 
 work for
 Voldermort types:
 typeid(a).to!string produces "std.range.iota!(int, 
 int).iota.Result", which
 is not very useful as it can't be mixed in a program, being a 
 voldemort
 type, eg:
 mixin("std.range.iota!(int, int).iota.Result a;"); //compile 
 error

 However, the compiler could help reduce it to something as 
 simple as
 possible, eg:
 "typeof(iota(int.init))"
 here this would work:

 mixin("typeof(iota(int.init)) a;"); //works

 Is it possible to do that in a generic way, such that it hides 
 as much as
 possible the details of the expression (here, 
 "complex_expr_returning_3()"
 should be simplified to int.init)

 Or is there another trick I could use to instantiate a variable 
 with same type as a?


 NOTE:
 typeof(a) a2;
 works even though
 http://www.drdobbs.com/cpp/voldemort-types-in-d/232901591?pgno=2 says it
 won't work, "Sorry, that won't work, the compiler will not 
 allow a
 Voldemort Type to be instantiated outside of its scope "
 (although that's doesn't help with my original problem)

It's possibly not a real Voldemort type, i.e. one that closes 
over a local variable and therefore needs a context pointer. DMD 
correctly rejects this code:

     auto foo(int a) {
         struct S {
             int bar() {
                 return a;
             }
         }
         return S();
     }

     void main() {
         auto a = foo(42);
         typeof(a) b; // Error: cannot access frame pointer of 
xx.foo.S
     }