It seems that Ned Konz wrote:
>> I took that approach, doing a quick hack that seems to work.
>> I modified #mouseDownInConnected: this way:
> [snip]
>
> I did the same thing, more or less, but added a check on the attached
> morph as well.

It can be nil, right?

> The NCConstraintMorph can have an arbitrary number of inputs. However,
> no one has ever used it that way, and in NCLineEndConstraintMorph  I've
> defined #input as "inputs first", and that's used a lot in the  code.

Ok, I've fixed it.

I've noticed the existance of #firstInput, which does more or less the
same, but has no senders (in my image).

#firstInput uses "self inputs", which removes nil entries, while #input
uses the instance variable "inputs" directly, resulting in the possibility
of a nil response.

Nils are only possible given that this is a WeakArray or there is another
possibility?

>> One thing I don't like is to check for the index (1 or 2)...
>> perhaps two methods would be more Smalltalk-like ;-)
>
> Maybe. I'll think about doing this, and trying to maintain backwards
> compatibility. Part of the problem is that I haven't found good names
> for the different ends yet.

From/To, Source/Target, Start/Finish?
I suppose the problem is because those names give the idea of a directed
graph, which is not always the case.

Have a nice day,

German Morales