This gets into something I've been thinking about the past few of weeks. The slashdotting of TurboGears has made me realize that Python web frameworks have reached another level since the Py Web-Off in March. Then, people were lamenting the multiplicity of incompatible frameworks, and wishing for a single framework with batteries included to compete with Rails. Now, a new generation of frameworks have arisen (TurboGears, Django, Subway, Aquarium, and indirectly Paste) that do this, and many of them use existing software to an extent not previously seen. Ironically, we're still increasing the number of frameworks, but at least integration and code reuse is getting better. Also, the entire word "framework" seems to be evolving. Previously it meant anything that parses a URL and publishes a string to the web. Now it seems to mean an entire vertically-integrated package -- that's what users are clamoring for. The question for Quixote is, what does it want to be in this new era? One option is to cede the beginners' market to TurboGears, and my hunch is that's the best. Aquarium's author (JJ Behrens) has expressed willingness to do this if TG (or another) becomes the frontrunner, and wishes such a framework existed when he wrote Aquarium. We can position Quixote as "a toolkit for advanced users who want more freedom than TurboGears gives them". That would argue against investing in elaborate tools to create a Quixote "project" directory: that work should go into supporting the emerging beginners' framework instead. Which will probably be CherryPy-based, since both TG and Subway use CherryPy, and CherryPy has the best WSGI interface to Paste. This is not necessarily bad for Quixote. CherryPy is the closest to Quixote of any of the frameworks given its "map a directory to a class" structure. I don't know if Quixote influenced CherryPy in that regard but it's possible. The pieces for Quixote/Paste integration are there; they just need to be packaged up. Perhaps Quixote will find other ways to integrate with CherryPy/TurboGears applications. Also, I read an article recently (I think it's http://www.groovie.org/articles/2005/10/13/python-web-framework-niches but the site is down and neither the Google cache nor the Wayback machine have it) saying that we should stop working on frameworks and start working on libraries, so that applications can use than rather than being contained by them. I'm not totally convinced of this, but Quixote does have several autonomous APIs (PTL, htmltext, form library, html functions, etc) that should be marketed on equal par with the publisher. -- Mike Orror