At http://digitalmars.com/d/version.html

This example is wrong:

class Foo
{
int a, b;
debug:
int flag;
}

It should be:
class Foo
{
int a, b;
debug int flag;
}

- Dave

Dave wrote:
> At http://digitalmars.com/d/version.html
> 
> This example is wrong:
> 
> class Foo
> {
> int a, b;
> debug:
> int flag;
> }
> 
> It should be:
> class Foo
> {
> int a, b;
> debug int flag;
> }
> 

Forgive me, what is wrong with the example as it stands?  Its working valid D and illustrates the 'debug' attr-stmt, albeit in a cheesy NoOp sort of way.

-- Chris Sauls

On Sat, 10 Sep 2005 04:26:02 -0500, Chris Sauls wrote:

> Dave wrote:
>> At http://digitalmars.com/d/version.html
>> 
>> This example is wrong:
>> 
>> class Foo
>> {
>> int a, b;
>> debug:
>> int flag;
>> }
>> 
>> It should be:
>> class Foo
>> {
>> int a, b;
>> debug int flag;
>> }
>> 
> 
> Forgive me, what is wrong with the example as it stands?  Its working valid D and illustrates the 'debug' attr-stmt, albeit in a cheesy NoOp sort of way.
> 
However, this fails ...

class Foo
{
  debug:
   int flag1;
  this()
  {
  debug:  //  found ':' instead of statement
   int flag2;
  }
}


-- 
Derek Parnell
Melbourne, Australia
10/09/2005 9:02:14 PM

In article <1j8s5arhhzgdn.urwa8lhaj35j$.dlg@40tude.net>, Derek Parnell says...
>
>On Sat, 10 Sep 2005 04:26:02 -0500, Chris Sauls wrote:
>
>> Dave wrote:
>>> At http://digitalmars.com/d/version.html
>>> 
>>> This example is wrong:
>>> 
>>> class Foo
>>> {
>>> int a, b;
>>> debug:
>>> int flag;
>>> }
>>> 
>>> It should be:
>>> class Foo
>>> {
>>> int a, b;
>>> debug int flag;
>>> }
>>> 
>> 
>> Forgive me, what is wrong with the example as it stands?  Its working valid D and illustrates the 'debug' attr-stmt, albeit in a cheesy NoOp sort of way.
>> 
>However, this fails ...
>
>class Foo
>{
>  debug:
>   int flag1;
>  this()
>  {
>  debug:  //  found ':' instead of statement
>   int flag2;
>  }
>}
>

Yes, and:

- The BNF doesn't specify the ':' form.
- The ':' form is not consistent with how version(id), debug(id) and the other
compiler conditionals work.

So, I think the example is wrong, which is fine with me (rather than adding more fluff to the language and being inconsistent).

- Dave

Dave wrote:
> In article <1j8s5arhhzgdn.urwa8lhaj35j$.dlg@40tude.net>, Derek Parnell says...
> 
>>On Sat, 10 Sep 2005 04:26:02 -0500, Chris Sauls wrote:
>>
>>
>>>Dave wrote:
>>>
>>>>At http://digitalmars.com/d/version.html
>>>>
>>>>This example is wrong:
>>>>
>>>>class Foo
>>>>{
>>>>int a, b;
>>>>debug:
>>>>int flag;
>>>>}
>>>>
>>>>It should be:
>>>>class Foo
>>>>{
>>>>int a, b;
>>>>debug int flag;
>>>>}
>>>>
>>>
>>>Forgive me, what is wrong with the example as it stands?  Its working valid D and illustrates the 'debug' attr-stmt, albeit in a cheesy NoOp sort of way.
>>>
>>
>>However, this fails ...
>>
>>class Foo
>>{
>> debug:
>>  int flag1;
>> this()
>> {
>> debug:  //  found ':' instead of statement
>>  int flag2;
>> }
>>}
>>
> 
> 
> Yes, and:
> 
> - The BNF doesn't specify the ':' form.
> - The ':' form is not consistent with how version(id), debug(id) and the other
> compiler conditionals work.
> 
> So, I think the example is wrong, which is fine with me (rather than adding more
> fluff to the language and being inconsistent).
> 
> - Dave
> 
> 

It isn't inconsistant.  All the attributes and attr-statements (such as 'version' and 'debug') work with the ':' form, and here's the kicker, /outside of function bodies/. Personally I think even that is a needless restriction, since the ':' form means to perpetuate/propogate a given attribute until either the current scope ends or a contradictory attribute is reached (or in the case of attr-statements, the corresponding 'else' clause is reached).

For documentation, check out:
http://digitalmars.com/d/attribute.html

-- Chris Sauls

In article <dfvbnf$628$1@digitaldaemon.com>, Chris Sauls says...
>
>Dave wrote:
>> In article <1j8s5arhhzgdn.urwa8lhaj35j$.dlg@40tude.net>, Derek Parnell says...
>> 
>>>On Sat, 10 Sep 2005 04:26:02 -0500, Chris Sauls wrote:
>>>
>>>
>>>>Dave wrote:
>>>>
>>>>>At http://digitalmars.com/d/version.html
>>>>>
>>>>>This example is wrong:
>>>>>
>>>>>class Foo
>>>>>{
>>>>>int a, b;
>>>>>debug:
>>>>>int flag;
>>>>>}
>>>>>
>>>>>It should be:
>>>>>class Foo
>>>>>{
>>>>>int a, b;
>>>>>debug int flag;
>>>>>}
>>>>>
>>>>
>>>>Forgive me, what is wrong with the example as it stands?  Its working valid D and illustrates the 'debug' attr-stmt, albeit in a cheesy NoOp sort of way.
>>>>
>>>
>>>However, this fails ...
>>>
>>>class Foo
>>>{
>>> debug:
>>>  int flag1;
>>> this()
>>> {
>>> debug:  //  found ':' instead of statement
>>>  int flag2;
>>> }
>>>}
>>>
>> 
>> 
>> Yes, and:
>> 
>> - The BNF doesn't specify the ':' form.
>> - The ':' form is not consistent with how version(id), debug(id) and the other
>> compiler conditionals work.
>> 
>> So, I think the example is wrong, which is fine with me (rather than adding more fluff to the language and being inconsistent).
>> 
>> - Dave
>> 
>> 
>
>It isn't inconsistant.  All the attributes and attr-statements (such as 'version' and 'debug') work with the ':' form, and here's the kicker, /outside of function bodies/. Personally I think even that is a needless restriction, since the ':' form means to perpetuate/propogate a given attribute until either the current scope ends or a contradictory attribute is reached (or in the case of attr-statements, the corresponding 'else' clause is reached).
>
>For documentation, check out: http://digitalmars.com/d/attribute.html
>
>-- Chris Sauls

It doesn't work (v0.131) with either version or debug and the BNF specific to debug and version doesn't specify it so either that example is wrong or the docs. are inconsistent and the compiler is wrong.

- Dave