Fetching contributors…
Cannot retrieve contributors at this time
62 lines (48 sloc) 3.18 KB
Project JMongo
The Mongo project provides native bindings for most popular languages. For the ruby language,
the project has a pure ruby execution path and a C extension to handle some of the heavy
lifting. Unfortunately, the performance of the pure ruby path under jruby is less than
stellar. Additionally, the C extension isn't compatible with jruby so we aren't able to take
advantage of any native code boost.
JMongo solves this problem by putting a thin ruby wrapper around the 10gen mongo-java-driver. The
goal is to provide a drop-in replacement for the mongo and bson gems along with complete
API compatibility.
The repo was was forked from Chuck Remes's (now deleted) repo.
% gem install jmongo
* Use jruby with 1.9 compatibility turned on JRUBY_OPTS='--1.9'
Almost all of the Ruby driver API is implemented
The the Ruby driver tests have been brought over and converted to be MiniTest based and
the collection and cursor test suites pass. NOTE: a few (2/3) tests have been skipped, you should look at
them to see if they affect you.
The Mongoid rspec functional suite runs 2607 examples with 28 failures when using JMongo
My Mongoid repo was forked after this commit (so newer funtionality/specs will be missing)
commit 6cc97092bc10535b8b65647a3d14b10ca1b94c8c
Author: Bernerd Schaefer <>
Date: Tue Jun 28 12:59:34 2011 +0200
The failures are classed in this way:
* Different Exception class being raised for BSON invalid keys
* Managing Replica Sets directly
* Managing Connection Pools directly
* XML serialization
* Ruby RegExp to BSON encode and decode (FIXED)
I will fix these problems in due course
Please note that the java driver handles the Replica Sets and connection pools
If you are using Replica Sets and want to use JMongo you should be OK if you use a URI to connect.
JMongo lets the Java driver handle reading from slaves and writing to master, although YMMV as I have not
tested JMongo with Replica Sets yet.
If you intend to use the fsync=true uri option to imply safe=true on your queries, at the moment you will also
need to specify the w option in the uri. e.g. mongodb://;w=1;
I have added non-blocking support to the Cursor if it is tailable and bound to a Capped Collection.
This is to have similar behaviour to the ruby driver as far as test expectations.
The unblocking behaviour is achieved by inserting a "poison doc" into the capped collection in a timeout thread.
In the options hash if you set await_data to hash, float or true it will not block on
a value of true will use the defaults, a value of float will set the timeout whaile a hash will allow you to
control the timeout as well as your own poison doc and the poison doc equality (lambda/proc) mechanism.
The default timeout is 0.125 seconds.
Based on the tailable cursor work above, I introduced a new feature unique to the driver: monitorable
collections. A monitorable collection, when inserted (atm) will also insert to a capped collection called a monitor. You can then subscribe to the monitor by giving the subscribe method two callbacks and a timeout period