Question

When using Google App Engine with Django-nonrel, is there any way to take advantage of the Async Datastore API when I declare my model classes with the Django API?

Answer 1

Ok, I’ve investigated a bit more and found an alternative way to deal with having a Django Model (i.e.: all Django features there) and still having access to the async API…

Mainly, using the datastore directly:

from google.appengine.api import datastore

and I already had methods to convert all my models to/from a json dict, so, it was mostly a matter of discovering how Django-Nonrel did it behind the scenes:

E.g.:

Considering a ‘Project’ class with to_json and from_json methods (i.e.: create from a dictionary)

For doing a simple query (it seems Run() will do the work asynchronously, so, one could do the query.Run() and later start another query.Run() and both would work at the same time):

query = datastore.Query(Project._meta.db_table)
for p in query.Run():
    p['id'] = c.key().id() #Convert from app engine key
    print Project.from_json(p)

Now, using the API to get an object asynchronously:

from djangoappengine.db.compiler import create_key
async = datastore.GetAsync(create_key(Project._meta.db_table, project_id))
p = async.get_result()
p['id'] = c.key().id() #Convert from app engine key
print Project.from_json(p)

So, it’s possible to retain the model with the Django structure and when needed some wrappers do the needed work asynchronously.