Commits on May 1, 2012
  1. Include dist in -Pwfk.

    bobmcwhirter committed May 1, 2012
Commits on Apr 30, 2012
  1. [TORQUE-794] Set ENV['TORQUEBOX_CONTEXT'] for all Ruby runtimes.

    lance committed Apr 30, 2012
    In the web runtime, it is set to 'web'.
    In the jobs runtime, it is set to 'jobs'.
    In the services runtime, it is set to 'services'.
  2. typo

    bbrowning committed Apr 30, 2012
  3. Get XA working on Oracle with activerecord-jdbc-adapter (TORQUE-567)

    bbrowning committed Apr 30, 2012
    The rails_transaction_spec passes for Rails 3.0 and 3.1 when tested
    against Oracle databases. There are a couple of caveats:
    * Only ojdbc6.jar is supported right now, not ojdbc5.jar
    * This has only been tested with activerecord-jdbc-adapter and not
  4. Make leaf-installation of components more targetted.

    bobmcwhirter committed Apr 30, 2012
    Do not install gems into JRuby if jruby is not included in the distribution (-Pwfk).
  5. Bump WFK version.

    bobmcwhirter committed Apr 30, 2012
Commits on Apr 29, 2012
  1. Rev stilts to fix infinite loop in TORQUE-787, processing WebSockets …

    bobmcwhirter committed Apr 29, 2012
    which are abbreviated and contain no newlines.
Commits on Apr 28, 2012
Commits on Apr 27, 2012
  1. Update messaging config to fix clustering on Linux platforms (TORQUE-…

    bbrowning committed Apr 27, 2012
    The config we shipped didn't properly cluster on many Linux and some
    Mac platforms so the move to <socket-bindings> was backported from AS
    7.1.2 to get things clustering as expected.
  2. Mynah Bird

    jcrossley3 committed Apr 27, 2012
  3. Do a better job of closing -knob.yml MountHandles (TORQUE-570)

    bbrowning committed Apr 27, 2012
    The structure of our KnobRootMountProcessor now more closely matches
    upstream AS7's DeploymentRootMountProcessor. However, the main bug fix
    is attaching our -knob.yml ResourceRoot as a new KNOB_ROOT attachment
    on the DeploymentUnit so that we can retrieve it later to unmount. We
    can't rely on unmounting the DEPLOYMENT_ROOT attachment because that
    gets changed in other processors to point to the actual app's root.
    The previous attempted fix for this just kept track of the -knob.yml
    MountHandle as a member variable but that would only work if the app
    being undeployed was also the last app deployed. Any other case would
    lead to the mount not getting unmounted and the error in TORQUE-570.
Commits on Apr 26, 2012
  1. Allow Rack apps to handle http OPTIONS requests (TORQUE-792)

    bbrowning committed Apr 26, 2012
    Instead of only executing Rack code if all other filters in the chain
    return an error, execute Rack code if other filters in the chain
    return an error or the response is still not committed. This lets
    OPTIONS requests get handled by HttpServlet and Rack applications so
    CORS (Cross-Origin Resource Sharing) headers can be set appropriately.
  2. Simplify newrelic background task monitoring.

    tobias committed Apr 26, 2012
    This has the side effect of fixing TORQUE-780, since we were requiring the
    newrelic_rpm gem before the pwd of the runtime was set by an initializer,
    so it couldn't find the app's config. We now only enable background tracing
    if the rpm gem has already been loaded, which will be the case if it's actually
    being used in the app.
  3. Incorporating feedback

    jcrossley3 committed Apr 26, 2012
  4. More complete logging docs

    jcrossley3 committed Apr 26, 2012
Commits on Apr 25, 2012
  1. Typo.

    tobias committed Apr 20, 2012
Commits on Apr 24, 2012
  1. Reverting "Conditionally pend all the integs that misbehave in domain…

    jcrossley3 committed Apr 24, 2012
    … mode"
    Revert "Conditionally pend all the integs that misbehave in domain mode"
    This reverts commit 65e8dc5.
    These don't make sense anymore since only integs in spec-domain/ are
    run in domain mode.
Commits on Apr 23, 2012
  1. Merge pull request #92 from qmx/docs_minor

    tobias committed Apr 23, 2012
    gcj. gcj is a little confuse
  2. gcj. gcj is a little confuse

    qmx committed Apr 23, 2012
Commits on Apr 19, 2012
  1. Don't require torquebox-web since RackRuntimeInitializer does (TORQUE…

    bbrowning committed Apr 18, 2012
    The require is harmless under 1.8 but due to a bug in JRuby's 1.9
    LoadService implementation (JRUBY-6610) this throws a LoadError if
    Bundler is used and torquebox or torquebox-web are not in the app's
    Gemfile. Since torquebox-web is already required by
    RackRuntimeInitializer, this was a no-op anyway.
  2. Can't inject messaging destinations without torquebox-messaging avail…

    bbrowning committed Apr 18, 2012
    …able (TORQUE-775)
    JRuby's LoadService is more strict and correct, in this case. This
    shouldn't have worked before but it did. You always need the torquebox
    or torquebox-messaging gem in your Gemfile to inject messaging
    destinations, since they call code inside the TorqueBox::Messaging module.
  3. Use reflection in ScriptAnalyzer so it works under JRuby 1.7 (TORQUE-…

    bbrowning committed Apr 17, 2012
    We should figure out why LocalStaticScope's constructor is no longer
    public in JRuby 1.7 and either adjust our code or submit a patch to
    make it public again.