This is a compiler bug. To allow usage of AA functions from druntime,
properties of built-in associative arrays are forwarded to the template
instance AssociativeArray!(K, V). This is erroneously also done for .init
and .stringof.
There are some bug reports for this.
http://d.puremagic.com/issues/show_bug.cgi?id=5683
If it were only init/stringof/mangleof this can be easily fixed.
On Thu, 28 Jul 2011 23:55:45 +0200, Jens Mueller <[email protected]>
wrote:
Hi,
I'm trying to compare the type of associative arrays:
unittest {
import std.stdio;
double[string] aa;
writeln(typeof(aa).stringof);
writeln(typeof(aa.init).stringof);
writeln(typeid(aa));
writeln(typeid(aa.init));
assert(is(typeof(aa) == typeof(aa.init)));
}
The assert fails which puzzles me. Why is the type of the expression aa
not the same as the type of the expression aa.init? There is a struct
AssociativeArray!(key, value) involved (part of druntime). The typeid
(i.e. the TypeInfo of the respective types of the expressions) are not
the same (double[immutable(char)[]] vs.
object.AssociativeArray!(string,double).AssociativeArray).
Jens
