In article <CqA56H.3rG@pau.synnet.com> eric@synnet.com (Eric Baur) writes:
>According to both the ARM and (at least one version of) the draft ANSI
>standard the following is NOT legal:

 There is no draft, and its a joint ANSI/ISO effort.
>
>    class X {
>         const int xdata = 100;
>    };
>
>Yet rumors persist that at the last meeting of the X3J16 committee
>this syntax for initialization of const members was "approved".

 Both X3J16 and WG21 approved this extension, yes.  But it is:

    class X {
         static const int xdata = 100;
    };

>Is this, in fact, the case?

 Yes, its the case.

>If so, were any additional semantics
>specified, e.g. are such members implicitly static?

 You have to specify 'static' I believe. Anything else
would be confusing. Non-static member may not be initialised in-class
(unfortunately)

--
        JOHN (MAX) SKALLER,         INTERNET:maxtal@suphys.physics.su.oz.au
 Maxtal Pty Ltd,      CSERVE:10236.1703
        6 MacKay St ASHFIELD,     Mem: SA IT/9/22,SC22/WG21
        NSW 2131, AUSTRALIA

In article <CqA7My.3ux@borland.com> pete@genghis.interbase.borland.com (Pete Becker) writes:
> What was passed is in-class initialization of static const data
>members. You must still provide a separate definition, and either the
>declaration or the definition can have an initializer, but not both.

 Woops, I forgot to say that.

 And we both forgot to say that it only works for
const _integral_ types initialised by constant expressions.

 In other words, its much the same as using an enum,
except you have to provide a definition. I think I'll stick to enums.
Less typing :-)
--
        JOHN (MAX) SKALLER,         INTERNET:maxtal@suphys.physics.su.oz.au
 Maxtal Pty Ltd,      CSERVE:10236.1703
        6 MacKay St ASHFIELD,     Mem: SA IT/9/22,SC22/WG21
        NSW 2131, AUSTRALIA

According to both the ARM and (at least one version of) the draft ANSI
standard the following is NOT legal:

    class X {
         const int xdata = 100;
    };

Yet rumors persist that at the last meeting of the X3J16 committee
this syntax for initialization of const members was "approved".
Is this, in fact, the case? If so, were any additional semantics
specified, e.g. are such members implicitly static?

Thanks,

Eric Baur                   eric@synnet.com
Synernetics Inc.

In article <CqA56H.3rG@pau.synnet.com>, Eric Baur <eric@synnet.com> wrote:
>According to both the ARM and (at least one version of) the draft ANSI
>standard the following is NOT legal:
>
>    class X {
>         const int xdata = 100;
>    };
>
>Yet rumors persist that at the last meeting of the X3J16 committee
>this syntax for initialization of const members was "approved".
>Is this, in fact, the case? If so, were any additional semantics
>specified, e.g. are such members implicitly static?
>

 That initialization is not legal, and won't be. After all, just because
it's const, that doesn't mean that every instance of X should have the same
value.
 What was passed is in-class initialization of static const data
members. You must still provide a separate definition, and either the
declaration or the definition can have an initializer, but not both.

 class X
 {
 static const int Value = 3;
 };

 const int X::Value;

or

 class Y
 {
 static const int Value;
 };

 const int Y::Value = 3;

 -- Pete

>>>>> Eric Baur <eric@synnet.com> writes:

> According to both the ARM and (at least one version of) the draft ANSI
> standard the following is NOT legal:

>     class X {
>          const int xdata = 100;
>     };

> Yet rumors persist that at the last meeting of the X3J16 committee
> this syntax for initialization of const members was "approved".
> Is this, in fact, the case? If so, were any additional semantics
> specified, e.g. are such members implicitly static?

Yes.  Such members must be declared static AND defined in exactly one
place; they mimic other static data members, rather than file-scope static
consts.

foo.h:
class X {
 static const int xdata = 100;
};

foo.C:
const int X::xdata;

Jason