I was playing with what makes the Javists proud, but stumbled over this:

class M {
    this(byte a) { _a = a; }
    byte _a;
    byte a() { return _a; };

    static m = new class(4) M {
        this(byte a) { super(a); }
        override byte a() { return 3+_a; }
    };
}

Error: class test.M.__anonclass10 has forward references
Error: no constructor for __anonclass10

Is this a compiler bug? There's no reason why the above shouldn't compile, no?


But that's not all, when I mark my class immutable...

immutable class M {
    this(byte a) { _a = a; }
    byte _a;
    byte a() { return _a; };
}

void main() {
    auto m = new class(4) M {
        this(byte a) { super(a); }
        override byte a() { return 3+_a; }
    };
}

... the compiler says: cannot implicitly convert expression (this) of type immutable(M) to test.M.
Same story for const classes. Again, compiler bug?


Tomek

Dnia 12-12-2009 o 13:09:49 Tomek Sowiński <just@ask.me> napisał(a):

> Error: no constructor for __anonclass10

This one seems to be unrelated to anonymous stuff.

class M {
    this(byte a) { _a = a; }
    byte _a;
    byte a() { return _a; };

    static m = new M(3);
}

I get: Error: no constructor for M.


Tomek

Tomek Sowiñski wrote:
> Dnia 12-12-2009 o 13:09:49 Tomek Sowiñski <just@ask.me> napisa³(a):
> 
>> Error: no constructor for __anonclass10
> 
> This one seems to be unrelated to anonymous stuff.
> 
> class M {
>     this(byte a) { _a = a; }
>     byte _a;
>     byte a() { return _a; };
> 
>     static m = new M(3);
> }
> 
> I get: Error: no constructor for M.
> 
> 
> Tomek

Well for this one, you're doing it wrong. You want:

class M {
    this(byte a) { _a = a; }
    byte _a;
    byte a() { return _a; };

    static M m;

    static this() {
        m = new M(3);
    }
}

- --
My enormous talent is exceeded only by my outrageous laziness.
http://www.ssTk.co.uk

Dnia 12-12-2009 o 15:00:56 div0 <div0@users.sourceforge.net> napisał(a):

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Tomek Sowiñski wrote:
>> Dnia 12-12-2009 o 13:09:49 Tomek Sowiñski <just@ask.me> napisa³(a):
>>
>>> Error: no constructor for __anonclass10
>>
>> This one seems to be unrelated to anonymous stuff.
>>
>> class M {
>>     this(byte a) { _a = a; }
>>     byte _a;
>>     byte a() { return _a; };
>>
>>     static m = new M(3);
>> }
>>
>> I get: Error: no constructor for M.
>>
>>
>> Tomek
>
> Well for this one, you're doing it wrong. You want:
>
> class M {
>     this(byte a) { _a = a; }
>     byte _a;
>     byte a() { return _a; };
>
>     static M m;
>
>     static this() {
>         m = new M(3);
>     }
> }

So it's the same when I get the "non-constant expression..." error and have to put in static ctors? Then what's with the "no constructor for M"?


Tomek

Tomek Sowiński wrote:
> Dnia 12-12-2009 o 15:00:56 div0 <div0@users.sourceforge.net> napisał(a):
> 
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA1
>>
>> Tomek Sowiñski wrote:
>>> Dnia 12-12-2009 o 13:09:49 Tomek Sowiñski <just@ask.me> napisa³(a):
>>>
>>>> Error: no constructor for __anonclass10
>>>
>>> This one seems to be unrelated to anonymous stuff.
>>>
>>> class M {
>>>     this(byte a) { _a = a; }
>>>     byte _a;
>>>     byte a() { return _a; };
>>>
>>>     static m = new M(3);
>>> }
>>>
>>> I get: Error: no constructor for M.
>>>
>>>
>>> Tomek
>>
>> Well for this one, you're doing it wrong. You want:
>>
>> class M {
>>     this(byte a) { _a = a; }
>>     byte _a;
>>     byte a() { return _a; };
>>
>>     static M m;
>>
>>     static this() {
>>         m = new M(3);
>>     }
>> }
> 
> So it's the same when I get the "non-constant expression..." error and have to put in static ctors?

Yes I think.

> Then what's with the "no constructor for M"?

Don't know.

It's best not to read too much into DMDs error messages, they leave a lot to be desired.

- --
My enormous talent is exceeded only by my outrageous laziness.
http://www.ssTk.co.uk

Dnia 12-12-2009 o 13:09:49 Tomek Sowiński <just@ask.me> napisał(a):

> But that's not all, when I mark my class immutable...
>
> immutable class M {
>      this(byte a) { _a = a; }
>      byte _a;
>      byte a() { return _a; };
> }
>
> void main() {
>      auto m = new class(4) M {
>          this(byte a) { super(a); }
>          override byte a() { return 3+_a; }
>      };
> }
>
> ... the compiler says: cannot implicitly convert expression (this) of type immutable(M) to test.M.
> Same story for const classes. Again, compiler bug?

It's not only about anonymous classes. Take a look at this:

immutable class M {
    this(int a) { _a = a; }
    int _a;
    int a() { return _a; }
}

immutable class PodM : M {
    this(int a) { super(a); }
    override int a() { return 3+_a; }
}

void main() {
    auto m = new PodM(3);
}

The above still throws the same error:
Error: cannot implicitly convert expression (this) of type immutable(M) to test.M
Error: cannot implicitly convert expression (this) of type immutable(PodM) to test.PodM

Compiler bug I guess...


Tomek

Dnia 13-12-2009 o 00:44:56 Tomek Sowiński <just@ask.me> napisał(a):

> Dnia 12-12-2009 o 13:09:49 Tomek Sowiński <just@ask.me> napisał(a):
>
>> But that's not all, when I mark my class immutable...
>>
>> immutable class M {
>>      this(byte a) { _a = a; }
>>      byte _a;
>>      byte a() { return _a; };
>> }
>>
>> void main() {
>>      auto m = new class(4) M {
>>          this(byte a) { super(a); }
>>          override byte a() { return 3+_a; }
>>      };
>> }
>>
>> ... the compiler says: cannot implicitly convert expression (this) of type immutable(M) to test.M.
>> Same story for const classes. Again, compiler bug?
>
> It's not only about anonymous classes. Take a look at this:
>
> immutable class M {
>      this(int a) { _a = a; }
>      int _a;
>      int a() { return _a; }
> }
>
> immutable class PodM : M {
>      this(int a) { super(a); }
>      override int a() { return 3+_a; }
> }
>
> void main() {
>      auto m = new PodM(3);
> }
>
> The above still throws the same error:
> Error: cannot implicitly convert expression (this) of type immutable(M) to test.M
> Error: cannot implicitly convert expression (this) of type immutable(PodM) to test.PodM
>
> Compiler bug I guess...

Should've DAFS first...
http://d.puremagic.com/issues/show_bug.cgi?id=2610
What's disturbing is that this bug is nearly a year old... Could this be that immutable classes lay there virtually unusable for so long?


Tomek