Web scripting with Python (My 2nd post!)
Dave Cole
djc at object-craft.com.au
Fri Mar 8 01:28:45 EST 2002
>>>>> "Paul" == Paul Boddie <paul at boddie.net> writes:
Paul> Writing a Web application framework is one of those things that
Paul> looks easy enough at first - how hard can it be to do a CGI
Paul> wrapper after all? The problem is, after experiencing the low
Paul> cost of entry with Python, the tasks just get harder and consume
Paul> more of your effort than the application you were wanting to
Paul> write.
I couldn't agree more. While writing the toolkit is fun, it is not
until you try to use the toolkit for a real application that you
realise what needs to be done. That is when you start losing hair
bigtime.
Paul> How do you do user sessions properly? How do you authenticate
Paul> users? Some frameworks already provide these kinds of facilities
Paul> without dictating how your application is going to turn out.
The approach I took here was to leave it mostly up the the user.
Security is one of those things that you are never going to get right
for all users in a toolkit.
Paul> Starting out at a low level is exactly the kind of trap that too
Paul> many projects fall into, producing a few 0.0.x releases before
Paul> disappearing into the void.
I suspect that is when the developers realise the scope of the task
they have undertaken. I hate to think of the amount of time/money it
has cost us to develop our small toolkit.
Paul> And that's just desktop applications - the added amusements of
Paul> Web framework development include things like security, which
Paul> one might think is easy to address, but it's usually far better
Paul> to use trusted security techniques than think up schemes of your
Paul> own.
Amen.
Paul> I'll accept that some frameworks have a perspective that can
Paul> seem hard to share with the developers, and others are under
Paul> fairly major and continuous development, but with all the
Paul> existing Web frameworks out there, I'm starting to believe that
Paul> we don't really need many more, let alone ones which might never
Paul> approach getting finished.
I agree. The reason that we developed our toolkit was that we could
not find a serious alternative to Zope. We needed something that we
could hand over to a customer who would/could not grok Zope.
Albatross is the third generation of that toolkit. As long as our
company is operating, we will be maintaining and developing the
toolkit.
Paul> But don't let me stand in your way. ;-)
Even if the toolkit never sees the light of day, you will learn a lot
while trying to build it.
- Dave
--
http://www.object-craft.com.au
More information about the Python-list
mailing list