If you can filter the subset based on the key_name values (some kind
of prefix) you can use the method described in
http://djidjadji.appspot.com/2011-02-24-Near-optimal-key-ranges-for-GAE-mapreduce
It uses a custom _split_input_from_namespace() method in a
mapreduce.input_readers.DatastoreInputRe
It just sounds like what you need are long running background tasks. If you
expect this to take more than 10 minutes, you can either do task queue
chaining with query cursors, or look into one of the pipelining tools:
1. Fantasm: http://code.google.com/appengine/articles/fantasm.html
2. Pipeline A
OK thanks for letting me know. I am certainly looking forward to being able
to map over a subset of entities. It'll make my implementation of
broadcasting using the Channel API much nicer!
--
You received this message because you are subscribed to the Google Groups
"Google App Engine" group.
T
The functionality isn't implemented yet. Right now priority #1 is getting
reduce out the door.
Ikai Lan
Developer Programs Engineer, Google App Engine
Blog: http://googleappengine.blogspot.com
Twitter: http://twitter.com/app_engine
Reddit: http://www.reddit.com/r/appengine
On Fri, Apr 8, 2011 a
I see on http://code.google.com/p/appengine-mapreduce/wiki/UserGuidePython
the following statement under "Current Limitations":
Only full range scan is supported, i.e. it's impossible to scan a subset of
a particular entity kind.
Is this because it is technically impossible or just that the fun
