apropos of nothing... here are my comments.

For many moons, the mems-exchange people have produced high-quality
software with clean APIs: Quixote, SCGI, and Durus are the ones I
know the most about.  They have done essentially no advertising for
these applications, preferring a kind of stealth mode where they produce
good software that *they* can use and leave the rest of us to figure it
out.

SCGI is gaining some? traction as a WSGI-capable server now.  It could
easily be "marketed" as a clean alternative to FCGI.

Quixote (which consists of at least the Publisher interface,
form/widgets library, and PTL) has always had a low but steady level of
background "community" activity.  Personally I think this is because it
just works for many people, but also because it's only accessible to a
moderately educated set of Pythonistas.  (Try explaining Quixotic
programming to a new programmer sometime...)

Durus is relatively invisible in the Python community, I feel, but it's
known and some people are using it.

What's my point?  These tools are known to aficionados but are largely
invisible to the Python community as a whole, despite their immense
value.

I personally think that there is value in making these tools more
visible.  OTOH I am more than a bit of a dilettante when it comes to
this sort of thing.  My Quixote2 tutorial hasn't been updated in months,
because of time constraints; I'm only now putting additional effort into
it.  This is my main personal plan for boosting Quixote's visibility and
usability.

What could people do?

 * spend more time on www.quixote.ca correcting stuff.  It's seriously
        out of date in some respects, and some concentrated efforts
        would be useful there.

 * update the Quixote documentation, perhaps by making it into pure
        reference docs (epydoc, anyone?), and make sure that what's
        there is correct.  Right now, it's pretty scattered and not
        very useful unless you're already committed to Quixote,
        and even then I find the source code is accurate while the docs
        are not.

        See my recent widgets.txt thingamajig for an example of where
        effort could usefuly be applied.

 * testing.  There's virtually no testing of any kind in Quixote.
        Luckily the mems-exchange folk turn out fantastic code, so
        this hasn't been a huge problem -- but it's less than reassuring
        to have such a central package be so untested.  IMO.

        twill + code coverage testing would be a good way to do
        this IMO.

 * write blog entries and/or break out the useful packages.

        - SCGI is still largely unknown, despite being a fantastic and
                *clean* way to run things.

        - PTL is a well-kept secret.

        - It is pretty easy to rewrite CherryPy's handler in Quixote.  I
                started doing this back in May; it took me less than two
                hours to get most simple CherryPy examples working in
                one handler.  This sort of thing is a good demo of
                Quixote's power.

        - examples.

        - more examples.

        - even more examples.

None of these things are things that the mems-exchange people should do,
honestly.  It's up to the rest of us schlubs.

cheers,
--titus