[BEEPbuilders] Proposal for Getting Started
Fri, 25 Oct 2002 19:31:28 +1000
Content-Type: text/plain; charset=iso-8859-1
On Thu, Oct 24, 2002 at 10:35:37AM -0700, Gabe Wachob wrote:
> On 24 Oct 2002, Jonas Borgstr=F6m wrote:
> > Great initiative, the RoadRunner project would be glad to participate!
> > With a common interoperability test, each library author probably doesn=
> > have to write their own regression tests.
> > Should the interoperability test only contain tests for the "core" feat=
> > or will it also test the TLS, SASL, XML-RPC... profiles?
> I hadn't thought about it, but I suspect that all well known profiles are
> appropriate for interop testing. As for profiles in particular, it seems
> that the common "tuning" profiles (e.g. TLS, SASL) are the most important
> to test first.
> I think this effort should focus on core functionality first though.
I concur. A bare minimum of interoperability is support of BEEP frames
and the BEEP Management profile. You could conceivably write an application
on top of a BEEP implementation that only had these functions. Of course,
the authentication and encryption capabilities provided by SASL and TLS
add great value. I believe these should come next, in this order, as that
is their level of complexity.
Perhaps we could come up with suitable labels for each level of=20
interoperability? In BEEPy (my implementation) I'm starting from the
simplest stuff first, nailing that down and then progressing on to the
more complex profiles.
I've yet to check out the SF site, but my BEEPy regression test are
in CVS at:
test_listener.py is probably the most interesting for this discussion. It
currently tests most of the corner cases for basic BEEP framing:
MSG, RPY, ERR, ANS and NUL. I imagine that's the place to start for
interop testing? More formal test specs/procedures are probably required.
Anyhow, it's beer o'clock here in Oz, so I'm off to the pub!
"I'd think a dominatrix would give you a switching power supply."
-- Eric The Read in a.s.r
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org
-----END PGP SIGNATURE-----