Spec Changes?

[mark_harrison_uk2]

Morning, guys...

Can I make a few general points about Spec Changes?

1: This is absolutely the right forum in which to discuss such things.

2: If you are posting a suggestion for a change/extension to xAP, can
you give thought (and indicate) one key point - whether this ACTUALLY
requires a change to the core 1.2 protocol, or whether this could be
implemented as a policy layer / schema on top of 1.2

3: Please _don't_ go and change your code to meet a proposal here. We
need a general consensus and agreement from everyone before we
actually make changes to the spec.

We discovered, in the process of putting together xAP 1.2 that a lot
of things had subtle wrinkles that we only thought of 2-3 days after
we'd initially discussed them, and had to pull back several times
from things that had initially looked interesting, but subsequently
proved to be flawed.

4: We have had problems getting some people to commit to xAP
because "it's always changing". As such, we need a clear roadmap
as
to what the Spec revision versions will be. This ties in with
point "2" above - anything that can be implemented as an OPTIONAL
policy-layer/addon rather than a core spec change is good.

5: Schemas, schemas, schemas.... Let's have a few more schema
proposals posted here. The more schemas we can formally adopt the
better.

Now, here's something contentious and should be viewed as a
suggestion: A schema could mandate certain constraints. For example,
any message in schema xap.x10 could REQUIRE that all message blocks
had unique names... and/or that there was only single message
block... and/or that name/value pairs had to have unique names within
a message block!

M.