[PEAK] Intended behavior of StructuralFeature._setup ?

[Bob Ippolito]

On Feb 24, 2004, at 12:54 AM, Bob Ippolito wrote:

> I'm still studying StructuralFeature here to try and figure out how 
> best to add support for mappings (first as a subclass in my own code, 
> then perhaps as a patch)...
>
> Right now I'm staring at the code for _setup, and it doesn't seem to 
> make a lot of sense:
>
> ...
>         if feature.isMany:
>             p = 0
>             value = tuple(map(normalize,value))
>
>             for v in value:
>                 doLink(element,value,p)
>                 p+=1
>
> note especially the doLink that uses "value" and not "v".  I assume 
> this is a bug?
>
> Also, what's the point of making a tuple of a map that you're just 
> going to iterate over?  Without using gc.get_referrers or similar 
> "fun" code, I can't imagine what could possibly change that list 
> during iteration.

Ok ignore that.. it's late :)

> FWIW, I decided to make a separate implementation of unlink for 
> mappings that are or are not ordered (isMany uses the same 
> implementation).  I can't imagine what kind of invariant you're trying 
> to preserve in a collection that's marked as unordered by deleting it 
> in reverse arbitrary order :)