I just saw the intent for this in the " split up compilation of the
setting" thread- that it is useful to:
>Enable a "class stub" syntax that allows us to declare a given symbol
> as being a valid class without having to declare the body of the
> class at that time. For example:
>
> class Rat { ... };
I can agree with that so long as the "yada" is the only token inside
the brackets. On the other hand why not go along with C convention and
allow
class Rat;
to pre-declare a class, or p5 convention
use class 'Rat';
-y
On Mon, Sep 7, 2009 at 9:44 AM, yary<not....@gmail.com> wrote:
> This spec subtly alters the meaning of "...". Whereas "yada" used to
> mean "this is not yet implemented, complain if executed" it now adds
> "but don't complain if it is a class fully implemented elsewhere".
>
> Allowing two implementations of a class iff one of them has a yada
> opens up maintenance issues. There's "action at a distance", when
> reading the source looking for the definition of "class Foo"- the
> reader could be looking at one and not seeing the other. Editors and
> code refactoring gets more complicated. It's too easy for two people
> to edit the different "class Foo"s without realizing there are two,
> especially if they leave the "yada" in place in anticipation of
> further editing & reviewing.
>
> Also this spec leaves some cases unspecc'ed. Does it matter which
> order the "yada" class and "complete" class are compiled, or if there
> are many "yada" classes? Do any of the definitions in the "yada" class
> carry over into the "complete" class?
>
> Does the same rule hold for methods, subs, all other named blocks, so
> we can put a "yada" in one to pseudo-comment it out?
>
> -y
>
>
>
>
> On Sun, Sep 6, 2009 at 11:48 PM, <pugs-comm...@feather.perl6.nl> wrote:
>> Author: moritz
>> Date: 2009-09-07 08:48:34 +0200 (Mon, 07 Sep 2009)
>> New Revision: 28196
>>
>> Modified:
>> docs/Perl6/Spec/S12-objects.pod
>> Log:
>> [S12] spec behaviour of stubbed classes
>>
>> This is a bit more general than what I had in mind first. If the implementors
>> say this is too hard to do it this way, we can degrade it to allow only a
>> single
>> '...' term in the body of a stubbed class.
>>
>> Modified: docs/Perl6/Spec/S12-objects.pod
>> ===================================================================
>> --- docs/Perl6/Spec/S12-objects.pod 2009-09-06 15:06:11 UTC (rev 28195)
>> +++ docs/Perl6/Spec/S12-objects.pod 2009-09-07 06:48:34 UTC (rev 28196)
>> @@ -55,6 +55,22 @@
>> class is also a module, it also handles any module-oriented keywords.
>> You can export subs from a class at "use" time, for instance.)
>>
>> +If the class body throws an exception from a literal C<...> (yada) term,
>> +the class defintion is considered incomplete, and a second definition of
>> +that class does not complain, so you can write
>> +
>> + class Foo {
>> + has $!some_attr;
>> +
>> + ... # literal ... here interrrupts class definition
>> + }
>> +
>> + # other code here
>> +
>> + class Foo {
>> + # rest of class defintion here
>> + }
>> +
>> A named class declaration can occur as part of an expression, just like
>> named subroutine declarations.
>>
>>
>>
>
