I’m trying to decide what host to use for a new Django project and would love to use the free GAE while its young, but this could be a deal breaker.

I’m a bit concerned about refactoring the data layer for my application as well, in case I chose to migrate to a non GAE platform. But the way I see it, having chosen GAE you have basically chosen a hierarchical key-value data store over a traditional SQL engine. Therefore I’m assuming that if you chose to port your application to using CouchDb, MongoDb, Dynamo, or Tokyo Cabinet (none of which I have used btw, BigTable being my cherry breaker away from usual SQL), it should not involve a massive rewrite. As opposed to say porting your app to using Postgres or Mysql.

What are your thoughts on that? Have you used any of the other key value stores or ported apps between them?

cheers

But the real trouble is django is about 800 files (12 megabytes) and for GAE the number of files count.

Also, apparently Django in GAE uses a extra amount of resources.

The helper is indeed a very nice tool, although it still makes you learn django (if you haven’t)

As for Guido, we all know he has a think for django.

One point against django that you forgot is that it’s made as a monolithic component, so ones you start using parts of it things get nasty, for example take a look at how much trouble the appengine people had to go thru to be able to use django-templates in webapp. Just read the module docstring of google_appengine/google/appengine/ext/webapp/template.py

That said I totally agree on the wsgi comment from Noah. In fact I can tell you that I’m using pylons on appengine, but the core is webapp and I don’t have pylons installed on that virtualenv :)

I’d love to hear more about the book you’re writing. If you need a tech reviewer, etc., give me a ping — I’m interested! :)

It is interesting the Guido is using Django himself for Rietveld.

I am currently in the middle of writing a book on Google App Engine, was at the Google I/O question, asked Guido in the Fireside chat about support for 3rd party components such as Django, and interpreted things in a different way.

I think the webapp framework IS the best approach at this moment. The real beautify of Python web development is the WSGI standard. There are many wonderful things happening with WSGI and the webapp framework currently supports WSGI fully unlike the current trunk version of Django. If you want to use Django components then it might make more sense to use the current django templates and URL dispatch that come with the webapp framework. In reality, webapp is a forked version of Django with support for WSGI, due to the brilliant work of Ian Bicking: http://ianbicking.com.

Thanks so much for reporting in — happy to know I made the right decision! :) And thanks also for the session notes.

Given that, I think it’s clear that the vast majority of App Engine development will happen with Django.

I wrote up my notes from the App Engine sessions at Google IO here: http://offermann.us/2008/05/29/day-one-at-google-io/