Cascade

This is some rough work on playing on the chainable Map/Reduce implementation for CouchDB. Its missing some important bits that will let you trash things

pretty easy but I figure its better than nothing.

TODO

• Add a gen_server to track in progress updates similar to the view pattern. Right now simultaneous clients will end up stomping on each other.
• Add status caching. Right now it still does a complete index scan when propogating updates unconditionally. Most likely gonna solve this with a _local doc on the root node.
• Add arbitrary barnching of work flows. Right now each _design doc is limited to a linear work flow to make the initial implementation easier.
• Add merging. I really want to add the merge phase from Map/Reduce/Merge.
• Better documentation...

Building

$ make

If that fails, well, you got this far...

Installing

No real installation yet.

Running

I run this with the CouchDB dev server runner.

$ cd $PATH_TO_COUCHDB
$ ./bootstrap && ./configure && make dev
$ ERL_LIBS=/path/to/cascade ./utils/run

Configuring

Add the following to your local_dev.ini:

[httpd_db_handlers]
_cascade = {cascade, handle_cascade_req}

Testing

$ cd ./test
$ ./curl.sh

This requires couchdb-python cause I was lazy. And curl. And some other things. Hopefully the tests give you the right idea in the general work flow.