On Oct 16, 2007, at 5:14 AM, Ryan J. McDonough wrote:
>
>> A client API would be nice, although I'm not sure the annotation-
>> driven approach makes as much sense as on the server side. But
>> unless I'm mistaken, anything we'd come up with would have to be
>> created in a hurry, and I don't believe there's a good enough
>> chance something very good would be the result.
>>
>> My vote would be to stick to the original plan and postpone this
>> to another JSR.
>
> Heiko brings up a great point about the TCK. As we develop the TCK,
> we're more than likely going to start encountering some type of
> common means of calling the server side code. So one has to ask, if
> we have to build this stuff anyway in order to provide a TCK, why
> not think about formalizing the effort as well?
>
Its a good point about the TCK and I agree that we'll get some good
experience building that. However, I'd prefer that we see how that
code turns out before we add a client API as a formal deliverable of
the JSR. I propose we add an issue re a client API and plan to take a
look at the TCK test client code once we make some progress on that.
Sound reasonable ?
Marc.
---
Marc Hadley <marc.hadley at sun.com>
CTO Office, Sun Microsystems.