Just a quick note: Darrell, your objection is noted. Paul, your reply 
to Darrell is noted, along with your objection to the Staged Context 
idea for A5. Stephen, I have noted your correction to the summary 
(context entries can be implied by the requested interface). I'll 
include all that in the next summary, which I intend to write after 
work today (say about 12hrs from now).

A correction: When summarizing the casting of a context interface,
I forgot to note that there are two ways being proposed for this:

 1. The component specifies an implementation class C that is
instantiated
    via the C(Map, Context) constructor. This instance is then passed
    to the component.

 2. The component specifies an interface I that it expects. The
container
    then uses its own context implementation class and dynamic proxies
    to provide the component with a context object.

I'll address this more in-depth in the next summary.

A request: Please don't use this thread to argue the technical merits
of each proposal. (So far I haven't seen anything like that, but just in
case...) I'm trying to simply list everything that's been put
forward with the purpose that no one should have to worry about their
suggestion or objection being forgotten. Do comment if you wish to
correct me regarding how the idea is presented, or regarding concensus,
though. 

Finally: I think there is enough to start going toward a proposal and
then a vote regarding Context. By that I mean that I'll try to summarize
what we can agree on, see if that is enough to vote on (i.e. the issues
we vote on are separate enough that we do not affect anything there
isn't
concensus on), and if so, write a proposal. We can then proceed to 
discuss that one - followed by some more summaries like the previous
one,
and so on, until the vote is a mere formality.

/LS


--
To unsubscribe, e-mail:   <mailto:[EMAIL PROTECTED]>
For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>

Reply via email to