A first message?

[Bret <mersault@busboy.sped.ukans.edu>]

	Well, since I haven't seen any traffic on here yet, guess I can start
things off.
	First, I think we're probably ready to begin the more detailed planning
phases; most of what we've talked about thus far has been on a rather high
level.  I'm not saying we should have half the server written in a day or
anything, but we could begin talking about stuff in more detail.  So, to
stir up discussion, here goes...

	1. What type of process model do we want to use?  I sent out a proposal
some time ago and haven't heard anything back... do we want to go with
that, what Apache uses, or something completely different?

	2. What support libraries, if any, are we going to use?  Dean's work with
respect to Apache and Netscape's NSPR library sound promising... maybe
something like that would provide a relatively stable cross-platform thread
abstraction?
	What about the ACE classes?  I'm afraid I haven't dug into these in depth,
but the mailing list indicates to me that they might be a bit of overkill
for what we're trying to do.  Still, I've been picking up instructive
little hints along the way... :)

	3. What kind of requirements do we want to impose upon our users?  ACE
requires egcs and some kernel patches to compile under Linux, at least the
4.4.x version did (and I assume 4.5 does as well).  It's know that STL
doesn't compile and/or work on a couple platforms.  gcc 2.8.x is apparently
quite buggy and _still_ doesn't handle templates correctly.  We've already
said these limitations are acceptable, but we need to be very clear about
what exactly those limitations are.

	4. What sort of module interface and "hooks" should we provide?  Will a
generic Module class be enough?

	Second, while we're discussing some of these issues, it would be prudent
to augment this mailing list with a Web page of some kind.  Maybe just
basic design issues and stuff, but at least something we can all see and
talk about.  We aren't at the stage where we need a CVS repository yet,
although if you want to set one up, Ben, have at it. :)
	Third, the step above would be useful in attempting to draw more voices
into the project, something I think we really need.  Not to say that any of
the three of us are unqualified or anything but, speaking for myself,  the
more people we have listening to the stupid things I say, the better our
chances of stopping me before I actually _do_ something stupid. :)
	Ben said he would mention the project in new-httpd.  Any other suggestions
about possible locations to recruit faces?
	Finally, I would like to return to a high-level discussion for a moment
and restate something I hope we're all in agreement with.  As I see it, our
primary design goals are, in decreasing order...

	- Correctness
	- Reliability
	- Performance
	- Portability

	Agreed? :)
	Think that's all, discussion is not only welcomed but encouraged... 
	Later...

								- Bret -