Bug or feature?

[Daniel Kozák via Digitalmars-d-learn]

On Mon, 29 Jun 2015 05:04:36 -0700
Jonathan M Davis via Digitalmars-d-learn
<digitalmars-d-learn at puremagic.com> wrote:

> On Sunday, June 28, 2015 11:37:59 Jack Applegame via
> Digitalmars-d-learn wrote:
> > I don't see any reason why it should not compile.
> >
> > import std.array;
> > import std.range;
> > import std.algorithm;
> >
> > class Foo {
> > }
> >
> > void main() {
> >   auto result = iota(3).map!(i => new immutable Foo).array();
> > }
> >
> > /usr/include/dmd/phobos/std/conv.d(4028): Error: cannot
> > implicitly convert expression (arg) of type immutable(Foo) to
> > test.Foo
> > /usr/include/dmd/phobos/std/conv.d(3931): Error: template
> > instance
> > std.conv.emplaceImpl!(immutable(Foo)).emplaceImpl!(immutable(Foo))
> > error
> > instantiating /usr/include/dmd/phobos/std/array.d(115):
> > instantiated from here: emplaceRef!(immutable(Foo), Foo,
> > immutable(Foo)) test.d(9):        instantiated from here:
> > array!(MapResult!(__lambda1, Result))
> 
> You haven't declared an immutable constructor, so you can't construct
> an immutable Foo. If the default constructor were pure, then it could
> be used to construct immutable objects even if it weren't immutable,
> but the default constructor is neither pure nor immutable, and purity
> isn't inferred for normal functions.
> 
> It might make sense as a feature request to request that a class'
> default constructor be pure (assuming that its base class constructor
> is pure), since theoretically, it should be able to be pure, but that
> would be a change in the language. What you're seeing is not a bug.
> It's how the current design works.
> 
> - Jonathan M Davis
> 

No it is a bug in std.conv.emplaceRef or more probably in std.array.