I'm trying to get a local copy of the data on the live appspot
application loaded into the dev_appserver environment on my local
machine.
I've downloaded the entire contents of openradar using the
included nukefile (nu rocks) and stored the resulting json to disk.
I've added api in main.py based on the already present uploading of
radars code to allow the upload of comments and written a little
python to upload the radars and comments into the locally running
dev server. All is working except threaded comments and
users.
One can't really get around the users problem on the local
machine as the set of users google is using is, well, the set of
users that google is using. That's not such a big deal for local
development (though I think I might have a way around that
brewing).
The comment.is_reply_to.key().id() that is exported when
downloading the json is the read-only unique ID of the unique key
that is assigned by app engine when the comment is created. It's
not possible as far as I can tell, without modifying the model, to
recreate comment threads when re-uploading data.
Possible solution: I can re-upload radars because they are
referenced by apple's radar numbers in the number attribute and not
by the google app engine unique ID. Perhaps we could create an
identifying unique (UUID perhaps?) attribute on comments that is
independent of comment.is_reply_to.key().id() so that threaded
comments can be downloaded and then re-uploaded with the threaded
is_reply_to relationships intact.
This will improve portability and make it easier for those who
wish to hack locally on some "real" data.
