website python engine (slashdot)

[Paul Boddie]

Sam Penrose wrote:
> 
> Everyone and their grandmother seems to be working on a slightly
> different Python web application/publishing environment. A few
> observations:
> 
> 1) It isn't hard. Heck, even I'm doing it.

Yes, it's a great spare time project. ;-)

> 2) Except for Zope, most of these projects seem to be one-man
> operations, despite the fact that the projects are mostly open source
> and the authors trying to recruit help.

I would agree, and this is why a few like-minded individuals got together on the
"python-web-modules" e-group:

  http://www.egroups.com/group/python-web-modules

Although some collaboration has taken place, many people have come forward with
modules of their own. This is related to your fourth point.

> 3) The first two points together suggest that Python (with the standard
> library included) is all the common code base anyone needs.

I would tend to disagree with that. Whilst it might be possible that people
could get by with writing Web frameworks using just the 'cgi' and 'Cookie'
modules, for example, many people would benefit from higher level modules. I do
accept that different people have different requirements from their Web
frameworks, and this does tend to cut down the amount of common code; for
example, some people like programs embedded in HTML pages, others prefer
template systems, and others prefer programmatically generated HTML.

However, the kind of higher level modules which would benefit framework
architects depends on the background and point of view of those architects and
how they believe they could be helped to write better frameworks more quickly
and effectively. There is some consensus on "CGI wrapper" programs, but how they
interact with application programs is not something everyone agrees on.

> 4) It also suggests that a year from now something very different will
> be going on. If all of us feel the need to work on this but none of us
> is quite satisfied with what anyone else is doing, it suggests that none
> of us is coming up with a very good answer. Since most of us are pretty
> smart (present writer excepted), that in turn suggests that we aren't
> going to get an answer until we reformulate the question. What are we
> missing here, folks? I have a hunch we're all going to be wishing we'd
> thought of it whenever it finally arrives...

I would say that what we are missing is a way of making Web applications which:

  * Is easy to buy into, well documented, "unsurprising".

  * Allows easy deployment for simple solutions. Permits more complicated
    deployment if that is necessary for performance, high-availability and so
    on.

  * Scales up to provide support for complicated interactions. (We don't really
    need another way of writing "one shot" HTML form handlers.)

  * Helps us manage the presentation and its dependency on the data structures
    being manipulated by application code.

  * Gives obvious solutions for issues like internationalisation and
    localisation, and many other commonly occuring issues. It isn't enough that
    a framework supports these things if only one can figure out how it is done
    - these things require clear documentation.

I am sure that I could think of many other requirements, but for me these are
already useful for assessing the projects out there today. My own spare time
effort advanced to the point of being a bit too difficult to buy into, but you
can find information about it, with links to some of the other projects, here:

  http://www.crosswinds.net/~pboddie/

Regards,

Paul