When I attempt to compile the following example code from the
section on UDAs
at:  http://dlang.org/attribute.html

void main(string args[]) {
template Tuple (T...)
{
         alias T Tuple;
}

enum EEE = 7;
@("hello") struct SSS { }
@(3) { @(4) @EEE @SSS int foo; }

alias Tuple!(__traits(getAttributes, foo)) TP;

pragma(msg, TP); // prints tuple(3, 4, 7, (SSS))
pragma(msg, TP[2]); // prints 7
}

I get the following errors:

UDAs.d(9): Error: basic type expected, not {
UDAs.d(9): Error: no identifier for declarator int
UDAs.d(9): Error: semicolon expected, not '{'

Line 9 is the one starting "@(3) ..."

Is this due to:
        a. a DMD bug (DMD64 D Compiler v2.063.2, Linux)
        b. a Documentation bug
        c. me not properly copying/pasting the example :/

Craig

P.S. This reCAPTCHA anti-spam think sucks.

On 2013-08-19 19:57, Craig Dillabaugh wrote:
> When I attempt to compile the following example code from the
> section on UDAs
> at:  http://dlang.org/attribute.html
>
> void main(string args[]) {
> template Tuple (T...)
> {
>           alias T Tuple;
> }
>
> enum EEE = 7;
> @("hello") struct SSS { }
> @(3) { @(4) @EEE @SSS int foo; }
>
> alias Tuple!(__traits(getAttributes, foo)) TP;
>
> pragma(msg, TP); // prints tuple(3, 4, 7, (SSS))
> pragma(msg, TP[2]); // prints 7
> }
>
> I get the following errors:
>
> UDAs.d(9): Error: basic type expected, not {
> UDAs.d(9): Error: no identifier for declarator int
> UDAs.d(9): Error: semicolon expected, not '{'
>
> Line 9 is the one starting "@(3) ..."
>
> Is this due to:
>          a. a DMD bug (DMD64 D Compiler v2.063.2, Linux)
>          b. a Documentation bug
>          c. me not properly copying/pasting the example :/
>

Move everything out of "main" and it will work.

-- 
/Jacob Carlborg

On Monday, 19 August 2013 at 18:01:40 UTC, Jacob Carlborg wrote:
> On 2013-08-19 19:57, Craig Dillabaugh wrote:
>> When I attempt to compile the following example code from the
>> section on UDAs
>> at:  http://dlang.org/attribute.html
>>
>> void main(string args[]) {
>> template Tuple (T...)
>> {
>>          alias T Tuple;
>> }
>>
>> enum EEE = 7;
>> @("hello") struct SSS { }
>> @(3) { @(4) @EEE @SSS int foo; }
>>
>> alias Tuple!(__traits(getAttributes, foo)) TP;
>>
>> pragma(msg, TP); // prints tuple(3, 4, 7, (SSS))
>> pragma(msg, TP[2]); // prints 7
>> }
>>
>> I get the following errors:
>>
>> UDAs.d(9): Error: basic type expected, not {
>> UDAs.d(9): Error: no identifier for declarator int
>> UDAs.d(9): Error: semicolon expected, not '{'
>>
>> Line 9 is the one starting "@(3) ..."
>>
>> Is this due to:
>>         a. a DMD bug (DMD64 D Compiler v2.063.2, Linux)
>>         b. a Documentation bug
>>         c. me not properly copying/pasting the example :/
>>
>
> Move everything out of "main" and it will work.

So I guess it was option (c) after all.

I am still curious though was is wrong using UDAs inside main().
If I have the following code inside main.

    @(7) int b;
    pragma(msg, __traits(getAttributes, b) );

or
    @("goodbye") struct D {}
    pragma(msg, __traits(getAttributes, D) );

Then it compiles fine.  But if I make it:

    @(7) { @{11} int b; }
    pragma(msg, __traits(getAttributes, b) );

Then I get the same error.

Craig

On Monday, 19 August 2013 at 18:21:26 UTC, Craig Dillabaugh wrote:
> I am still curious though was is wrong using UDAs inside main().

It is not exactly about UDA - key is different meaning of curly brace {} syntax in function bodies and in declaration scopes.

Used in declaration scope {} apply all attributes attached to it to every declaration inside:

const
{
    int a;
    double b;
}

static assert (is(typeof(a) == const));

This also applies for UDA's.

In function body it creates new variable scope - variables declared inside get destroyed upon leaving the scope and are not available outside.

void foo()
{
    {
       int a;
    }
    // a = 42; // error
}

As this usage of {} does not expect any attribute propagation, UDA's get rejected too.

On Monday, 19 August 2013 at 18:28:24 UTC, Dicebot wrote:
> On Monday, 19 August 2013 at 18:21:26 UTC, Craig Dillabaugh wrote:
>> I am still curious though was is wrong using UDAs inside main().
>
> It is not exactly about UDA - key is different meaning of curly brace {} syntax in function bodies and in declaration scopes.
>
> Used in declaration scope {} apply all attributes attached to it to every declaration inside:
>
> const
> {
>     int a;
>     double b;
> }
>
> static assert (is(typeof(a) == const));
>
> This also applies for UDA's.
>
> In function body it creates new variable scope - variables declared inside get destroyed upon leaving the scope and are not available outside.
>
> void foo()
> {
>     {
>        int a;
>     }
>     // a = 42; // error
> }
>
> As this usage of {} does not expect any attribute propagation, UDA's get rejected too.

Thank you, makes perfect sense.