There is a behaviour difference with the old NDB that we're relying on wrt to cursor_after() at the end of iterating through a query. Given the following: ```python class Model(ndb.Model): a = ndb.IntegerProperty() Model(a=1).put() Model(a=2).put() Model(a=3).put() result = Model.query().iter(limit=2) print(list(result)) print(list(Model.query().iter(start_cursor=result.cursor_after()))) ``` The old NDB gives: ``` [Model(key=Key('Model', 31), a=1), Model(key=Key('Model', 32), a=2)] [Model(key=Key('Model', 33), a=3)] ``` The new NDB gives: ``` [Model(key=Key(u'Model', 31L), a=1L), Model(key=Key(u'Model', 32L), a=2L)] BadArgumentError: There is no cursor currently ```
There is a behaviour difference with the old NDB that we're relying on wrt to cursor_after() at the end of iterating through a query.
Given the following:
The old NDB gives:
The new NDB gives: