Hi! I have followed with interest the wiki page and your discussion. As I'll have no chance of testing the frameworks in the next two weeks, so I thought I may share some thoughts here.
I worked extensively with Twisted and Genshi, both of them are nice tools and smart, Genshi had some learning curve, but Twisted was just awkward. The documentation was missing and very confusing (somehow, I tend to judge frameworks based on the quality of their documentation and especially API documentation -- because when they have a smart tools, they also get more spare time to write it..) and in short, it was unpleasant and tiresome work. Sadly, the cool projects are sometimes cool only for experienced programmers -- I imagine that now I would be very comfortable with Twisted, but it wasn't the case few years ago when I was learning Python.. It might be subjective, but I haven't seen in your discussion one very important element - a lot is said about speed and personal preferences for templating engines, but there were only slight hints on two developer virtues: laziness and curiosity. I missed the elements of joy and discovery. I strongly believe that the main role of web frameworks is not simply to have the job done, but to give developers chance to have it done in joyful way, to make a work a lot of fun and not dreadful boredom, to marvel at smartness of other guys and generally to have great deal of fun and satisfaction discovering new things. Twisted, for example, is the right tool if you are writing asynchronous aps and web protocols, streaming etc, but Pylons or Django give more satisfaction of the 'actual work done' and new services build. Which of the two directions apply to us? Well, it depends... yes, I know, it is subjective, some of us are just interested in flying the plane, others want to construct it as well... I think that with miniframeworks more of the components need to be constructed, and your smart minds should give deeper thoughts to the connection between the internals of the web framework and its joints with the templating layer (for example) and not solve everything cheaply by keeping the executable actions on the template side (are we always going to go with the cheapest solution?). There are certainly more mistakes that we can make, and it is not as simple as it might seem. That's why those smart guys keep rewriting their web frameworks, right? Hopefully you will have a good eyes. To get the toiling labourers some source of joy and moments of 'Eureka!'. To satisfy both your tendency to build the aircraft and the urge to fly it high! Cheers, roman On Mon, Oct 3, 2011 at 10:17 AM, Tibor Simko <[email protected]> wrote: > Hello: > > The next Invenio Developers Forum will take place today starting at > 16:30 CEST in the EVO videoconferencing meeting called `Invenio > Developers Forum' and in person at CERN in the videoconferencing room > 31-S-023. > > The topic for today is the continuation of musings about web frameworks > for Invenio: > > <http://invenio-software.org/wiki/Talk/WebFrameworks> > > aiming at discussing concrete pros/cons and shortlisting potential > candidates. > > Best regards > -- > Tibor Simko >
