I'm sitting here lurking on this thread and to me the proposal makes
everything seem convoluted. Though user??.website??.profile?? could
feasibly be worked out by a new to intermediate user as asking
progressively if user was present and if so was a website object associated
with that user and if so was a profile associated to the associated website
present. But the syntax of that semantic isn't really clear or 'natural'.
Its starting to feel like things are moving towards being overly terse
rather than clarity being a retainable goal.
On Apr 2, 2013 4:57 PM, "Matthew Kerwin" <matthew / kerwin.net.au> wrote:

> On Apr 3, 2013 12:03 AM, "rosenfeld (Rodrigo Rosenfeld Rosas)" <
> rr.rosas / gmail.com> wrote:
> >
> > I believe you didn't understand the proposal, Matthew. a??.b would mean
> (a.nil? ? nil : a.b). So user.profile??.website??.thumbnail is equivalent
> to:
> >
> > user.profile.nil? ? nil : (user.profile.website.nil? ? nil :
> user.profile.website.thumbnail)
>
> You are correct, I thought a.b?? meant (a.respond_to? :b ? a.b : nil)
> since I thought I saw earlier something like `foo.empty???бн
>
> In that case the missing question marks were earlier in the sequence:
>
>     if user??.profile??.website??.thumbnail
>