Webware for Python

Hello,

> The Webware sprint members at PyCon were Tripp Lilley (figurehead),
> Ian Bicking, George Paci and myself.

It's great to see this kind of thing going on, especially given Ian's recent
work on Application (which really needed some attention, in my opinion).

> Monday Tripp, Ian and I mostly talked about what we wanted to do.  We
> identified documentation, components, a short-term user manager with a
> few features, a long-term user manager with more features, and 
> spec'ing out a CMS (Content Management System).  We decided to work on
> documentation because that's the key to getting the other things done.

About the user manager and user-related stuff, I recently found myself
adding various convenience methods to HTTPRequest in order to get things
like language and preferred content type information out of the environment.
Such settings are not necessarily just trivial 'getenv's since the
information is formatted in a certain way. It might be interesting for me to
submit patches, but such functionality could legitimately belong in the
'cgi' standard module, too.

Moving on to user handling, though, such preference information is
reasonably important for the handling of anonymous users, and it could be
interesting to be able to treat anonymous users in the same way as
identified users. Under the covers, the identified users would have their
preferences stored in session objects or in another location, but the API
for accessing such information would remain the same. As a first step to a
more comprehensive user handling API, I think that such functionality could
be helpful and yet relatively uncontroversial.

[...]

> I also talked with Twisted's documentor.  They have a system called
> Twisted Lore that uses XHTML-like files as source and produces HTML
> files with TOCs and footnotes, PDF and LaTeX files.

An overview of Python documentation systems would be interesting. There
could be as many of those as there are Web frameworks.

[...]

> (2) Twisted Lore currently depends on a homegrown XML parser in
> Twisted, which I'm reluctant to depend on.

I may have misinterpreted the Twisted people here, but there was a thread on
the XML-SIG mailing list where one of them claimed that the DOM APIs weren't
nice enough (or Pythonic enough, as is usually claimed, resulting in the
production of yet another incompatible interpretation of Pythonic that
doesn't work with other XML tools), and I seem to recall that the XML-SIG
regulars weren't exactly impressed by their assertions. I'm sure that one of
the things that they also wanted to do was to implement a "forgiving" XML
parser, and that kind of thing is always going to receive criticism in the
XML community.

Anyway, this was some time ago, and I haven't been on a number of mailing
lists for ages. Still, it's good to see Webware moving along.

Paul