Hoping to get a comment from the GAE python team on this one.
Has something changed between 1.6.3, 1.6.4 with regards to the way the dev_appserver datastore is flush
It looks like there have been some changes. I've been able to hack around the problem with the following:
from google.appengine.tools import dev_appserver
import atexit
atexit.register(dev_appserver.TearDownStubs)
This ensures the datastore is flushed on exit.
Before 1.6.4, we saved the datastore after every write. This method does not work when simulating the transactional model found in the High Replication Datastore (you would lose the last couple writes). It is also horribly inefficient. We changed it so the datastore dev stub flushs all writes and saves it's state on shut down.
Following the code:
It looks like manage.py should work if the server is shut down cleanly (with a TERM signal or KeyInterrupt).
Thanks to dragonx for his solution and info. I run my devserver from eclipse, and I was amazed to see my data not beeing saved after upgrading to 1.6.4 I added a flush to the database after every web request, to do that I implemented a base class for all requests and override dispatch:
developmentServer = False
if os.environ.get('SERVER_SOFTWARE','').startswith('Development'):
developmentServer = True
class BaseRequestHandler(webapp2.RequestHandler):
def dispatch(self):
retValue = super(BaseRequestHandler, self).dispatch()
if developmentServer:
from google.appengine.tools import dev_appserver
dev_appserver.TearDownStubs()
return retValue
informing about a change in behavior like that in the release notes, would have saved me two days of searching what went wrong in my upgrade.