aboutsummaryrefslogblamecommitdiffstats
path: root/README.rdoc
blob: 15b909a1591eb47e24010056237b721dfc71144a (plain) (tree)

































                                                                                                                                
== About

This is Nick's connection pool branch, wherein he attempts to rewrite Rails' connection handling code to
be more thread-safe, and to add connection pooling features.

== Goals

- Preserve Rails' lazy connection acquisition and caching strategy behavior as it exists prior to this work.
- Add ability to configure a connection pool to limit the number of connections made to a database in multiple-thread scenarios.
- Threads will block for a configurable timeout if the pool is exhausted until a connection is available.
- If none is available during the timeout period, an exception will be raised.
- Add a checkout/checkin API to reserve and release a connection to/from the pool.
- Add several different connection handling/pooling classes to serve different needs:
  - proper fixed-size connection pool
  - connection-per-thread with no maximum on the number of connections
  - single thread cached connection
  - pass-through to external connection pool (JRuby/JNDI data source connection pool)

== TODO

Remaining tasks:

- Fixed-size connection pool
- Add checkin/checkout API
- Add a #with_connection API that allows checkin/checkout of a connection outside of a provided block.

    Model.with_connection do |conn|
      Thread.new {
        Model.connection = conn
        # do something with conn
      }
      Model.connection.select ....
    end
- Look at whether existing clear_* or verify_* methods can be deprecated or removed.