On Tue, May 21, 2013 at 2:40 PM, Frank Shearar <frank.shearar at gmail.com>wrote:

> allMethodsSelect: aBlock localTo: aClass
>     "Answer a SortedCollection of each methodr in, above, or below the
> given
>     class that, when used as the argument to aBlock, gives a true result."
>
>     | aSet |
>     aSet := Set new.
>     Cursor wait showWhile:
>         [aClass theNonMetaClass withAllSuperAndSubclassesDoGently:
>             [:class |
>                 class selectorsAndMethodsDo:
>                     [:aSelector :aMethod|
>                         (aBlock value: aMethod) ifTrue:
>                             [aSet add: (MethodReference new
> setStandardClass: class methodSymbol: aSelector)]]].
>         aClass theNonMetaClass class withAllSuperAndSubclassesDoGently:
>             [:class |
>                 class selectorsAndMethodsDo:
>                     [:aSelector :aMethod|
>                         (aBlock value: aMethod) ifTrue:
>                             [aSet add: (MethodReference new
> setStandardClass: class methodSymbol: aSelector)]]]].
>     ^aSet
>
> Um. That is just an accidental duplication of the aClass
> theNonMetaClass class withAllSuperAndSubclassesDoGently: block, right?
>

No, it's not :)  It's preserving existing behavior, which is to search
class and instance side.  Note that

 aClass theNonMetaClass withAllSuperAndSubclassesDoGently:
    [:class|
    {class. class class} do: aBlock

is not equivalent to

    aClass theNonMetaClass withAllSuperAndSubclassesDoGently: aBlock
    aClass theNonMetaClass class withAllSuperAndSubclassesDoGently: aBlock

because the latter searches Behavior, ClassDescription, Class etc, while
the former doesn't.


> frank
>
>


-- 
best,
Eliot
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.squeakfoundation.org/pipermail/squeak-dev/attachments/20130521/8b72b3ae/attachment.htm