http://d.puremagic.com/issues/show_bug.cgi?id=1619

           Summary: "Missing initializer for const field" only reported if
                    an explicit constructor is present
           Product: D
           Version: 1.022
          Platform: PC
        OS/Version: Windows
            Status: NEW
          Severity: normal
          Priority: P2
         Component: DMD
        AssignedTo: bugzilla@digitalmars.com
        ReportedBy: ary@esperanto.org.ar


The following code gives a compiler error:

---
class X {
    const int x;
    this() { }
}
---

main.d(3): constructor main.X.this missing initializer for const field x

But this one:
---
class X {
    const int x;
}
---

compiles correctly. Since the semantic of both files is the same, an error should also be reported in the second case.


-- 

http://d.puremagic.com/issues/show_bug.cgi?id=1619


Andrej Mitrovic <andrej.mitrovich@gmail.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |andrej.mitrovich@gmail.com


--- Comment #1 from Andrej Mitrovic <andrej.mitrovich@gmail.com> 2013-01-23 11:08:02 PST ---
I don't know whether this is a valid report. The same behavior is in D2, afaik the field will be initialized with .init, whereas if you introduce a constructor the compiler expects you to initialize it yourself.

Can anyone with more insight confirm?

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------

http://d.puremagic.com/issues/show_bug.cgi?id=1619


Maxim Fomin <maxim@maxim-fomin.ru> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |maxim@maxim-fomin.ru


--- Comment #2 from Maxim Fomin <maxim@maxim-fomin.ru> 2013-01-24 07:24:35 PST ---
(In reply to comment #1)
> I don't know whether this is a valid report. The same behavior is in D2, afaik the field will be initialized with .init, whereas if you introduce a constructor the compiler expects you to initialize it yourself.
> 
> Can anyone with more insight confirm?

In both cases i is initialized to 0.

import std.stdio;

class A
{
    const int i;
}

class B
{
    const int i;
    this() { writeln(i); i = 2; }
}

void main()
{
    A a = new A;
    B b = new B;
}

As I understand the point of the requirement is that after exiting from ctor, const member cannot be altered (this assumption can be invalidated). So, the only opportunity to make const member store useful runtime value is during constructor invocation. This ability cannot be replaced by enum or manifest constant because they are shared across all instances. So, the only opportunity to configure at rt a const member which can be different across instances is a ctor (this assumption can be broken).

However, loosing this opportunity is not an error : if a user forgot to set up a const member, he can easily go back and modify it. Also a const member is still initialized, so no invalid value is created.

When this feature was added it likely had a rationale, but I don't see why not initializing a const member is treated so severely as an error.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------