Experiments with command-line and web flow adaptor

[Cantor, Scott]

On 4/8/13 3:21 PM, "Tom Zeller" <tzeller at dragonacea.biz> wrote:

>>In fact, I think
>> BaseContext itself belongs there. MessagingContext, et al. are really
>>just
>> specific subtypes and classes for messaging that would rely on
>> -profile-api.
>
>How about opensaml-context-[api|impl] ? If not, why ? I do not have
>strong feelings here, but I have wondered for awhile, so the
>conversation is more important than the outcome.

My immediate reaction is just that it hints of something sort of security
related because the word context makes me think of that, but I don't know
that it's anything but a personal reaction.

The more on-point comment is just that I'd say we should focus on the
other question first. Is "ProfileRequestContext" the right name for the
thing that the actions are presumed to operate on? If so, it seems like
-profile is appropriate.

>Agreed. I consider most of that boilerplate to be hop-scotching nulls
>while navigating the context hierarchy. Maybe a useful pattern will
>fall out.

I think in part that really is the pattern. Any null checks are a sign
that something's missing from the LookupStrategy thingies. But I don't
know if it's going to get much simpler than:

FooContext = lookup1();
BarContext = lookup2();

Maybe this is a place where Optional<FooContext> works. That becomes the
way to express "could be missing" on the return side, so it's obvious from
the local variables which contexts are required or not.

FooContext foo = lookup1();
Optional<BarContext> bar = optlookup2();

-- Scott