Webware for Python

Ian Bicking [mailto:ianb@...] wrote:
>=20
> I agree -- I don't think a Webware book would work.  In fact, I don't=20
> think Python is ready for any web programming books (except Zope -- =
but=20
> even those suck tremendously, so maybe Zope isn't ready either ;).

There's a question of maturity and stability when it comes to writing =
books
about technology. I remember the thousands of Java titles when Java was
still "white hot", but it doesn't endear you to your readership when the
book they bought is out of date before they even get it.

> The reality is that Python web programming is a fucking mess.  It's=20
> pathetic.  There's no way to write a book, because there's no material =

> that's appropriate book material.

I'd agree with that. At the "application plus nice framework" level, =
there's
too much fluidity in the technologies - ie. people are collaborating
actively to get stuff working, but many of them are happy to follow CVS.
Even at the lowest levels, there's general dissatisfaction with the most
basic of technologies: how many people have been campaigning for the CGI =
and
HTTP stuff to be rewritten?

> I like Webware, but it's not going to take over the world, not even =
the
> modest Python world.  Neither will Zope, not even Zope 3... and if I
> thought it would, maybe I'd switch over happily.

All those people wanting to know what Python's equivalent of J2EE is =
should
just pick up Zope 3 when it's ready (or before, if they're old hands =
with
J2EE). Huge APIs, big specifications, unfathomable deployment =
descriptors -
it looks as if it's all there. ;-)

To make Python solutions more attractive to newcomers (and even Python
programmers who are new to Web application development), I really think =
we
need to separate out the main points of concern for any given Web
programming framework:

 * How easy it is to install/deploy.
 * The programming/development style.
 * The available applications and companion technologies.
 * Community experience/expertise.

Right now, the newcomer stands daunted in front of that list on the
PythonInfo WebProgramming page and has to work out what the tradeoffs =
are in
choosing a handful of those frameworks. If they choose Webware...

 * Can they install it easily enough? What if they don't know Apache? (A =
lot
   of people run screaming from Apache's httpd.conf and it's easy to =
forget
   that after a while.)
 * Do they have to use Page classes or PSP? How else can they build
   applications? Are there any good examples?
 * What integrates with Webware? It's all very well that there are a =
million
   and one template frameworks, but which ones don't only work with =
Twisted?
   If I see a nice Web application, will it work on Webware or will I =
end up
   installing mod_python as well?
 * Are there people out there who know the technologies inside out? And =
the
   usual question: which big sites run Webware?

I know people who would quite like to do more Web development in Python, =
but
it's hard to ease them into it when you can install out-of-the-box =
solutions
which run PHP applications, for example.

What I've been getting at with WebStack is the removal of the whole
installation issue from the selection of higher-level applications or
technologies. In other words, if you've seen a nice application and it =
works
with WebStack, you get to run the same application on BaseHTTPServer (if =
you
don't know how to set up Apache), Webware (if you're already running it) =
or
a number of other alternatives including CGI (if you're interested in =
low
end hosting). People don't need to be told that the application =
concerned
requires them to install Apache (or whatever) and do a load of other
seemingly unrelated things before they can run the thing.

One desirable consequence of better standardisation is that developers =
can
just move on to higher-level issues and stop reinventing the wheel at =
the
low (and uninteresting) end.  And one side-effect of WebStack is that it
ends up just like the WebProgramming page, but with the differences =
encoded
in working source code. ;-)

Paul