Permalink
Commits on Mar 11, 2011
  1. Removing stupid test.

    bobmcwhirter committed Mar 11, 2011
  2. Adjust rspec test to not delete Dir.tmpdir, since that's OS-owned, an…

    …d makes things cranky when you delete /tmp.
    
    Tests did not fail previously, I think due to bug in VFS path-handling around those tests themselves.
    We now scribble our own workdir underneath /tmp and go from there, leaving /tmp in-tact.
    
    For whatever value of /tmp your operating-system uses.
    bobmcwhirter committed Mar 11, 2011
  3. Set the default max-deploys-before-restart to 100 for arquillian.xml.

    Then, on Windows, set it to 10.
    bobmcwhirter committed Mar 11, 2011
  4. Enforce binary mode on our rewindable inputs to satisfy Rack's idea o…

    …f line-endings and JRuby's subversion of them, when dealing with multipart mime form POSTS.
    bobmcwhirter committed Mar 11, 2011
Commits on Mar 10, 2011
  1. Fix bootstrap on Windows by using getPath( path ) instead of getRoot(…

    …).getChild( path ) which breaks on Windows
    
    since root is already C:.
    
    Also, when deploying via -knob.yml, sanitize the root: parameter appropriately for Windows.
    
    When setting RubyAppMetaData#setRoot(), use the string-sanitizing versions.
    bobmcwhirter committed Mar 10, 2011
  2. Fix double-quoting issues in assembly pom. A regression in jruby-mave…

    …n-plugin, but workaroundable.
    bobmcwhirter committed Mar 10, 2011
  3. Allow build-numbering magic to not fail on Windows.

    The jruby-maven-plugin seems to have some issues with
    double-quotes.
    bobmcwhirter committed Mar 10, 2011
  4. Remove lingering debug.

    bobmcwhirter committed Mar 10, 2011
  5. Fixes for Windows, particularly around closing files we open during t…

    …ests, plus a bugfix within our own IO#read.
    bobmcwhirter committed Mar 10, 2011
  6. Reworked the way you pass options to each on destination, and actuall…

    …y used the options.
    tobias committed Mar 10, 2011
Commits on Mar 9, 2011
  1. Restarting cache correctly now when configuration changes.

    Data persists across redeployments if cache configuration does NOT change.
    jcrossley3 committed Mar 9, 2011
Commits on Mar 8, 2011
  1. Mention how to run tb

    jcrossley3 committed Mar 8, 2011
  2. Include dependencies for the torquebox-base.gem to slurp in Ininfispa…

    …n and her dependencies.
    
    Rename our torquebox-base.rb to gem_hook.rb, since gem-maven-plugin generates the torquebox-base.rb.
    bobmcwhirter committed Mar 8, 2011
  3. Breaking out portions of the Ruby code that had been lodged inside of…

    … Java JAR
    
    files into their own gems.
    
    Each component/X might now have an X-gem/ module to build its related RubyGem.
    
    Also renaming/relocating various useful ruby classes.
    
    TorqueBox::Kernel, for instance, is now defined in 'torquebox/kernel'.  Likewise,
    TorqueBox::ComponentManager lives in 'torquebox/component_manager'.
    
    When loading web session stores using symbols, the resulting determined class
    is different between Rails2 and Rails3.  At this point, stupid copy/paste of
    the ActionController::Session-based impl to ActionDispatch::Session.
    
    ** TODO: This should be refactored into a common (rack-based) base, with simply appropriate
    aliases to the names and files Rails2/Rails3 expects for its own blend of pickiness.
    bobmcwhirter committed Mar 8, 2011
Commits on Mar 7, 2011
  1. Move RubyApplicationDeployer to PRE_REAL to see if that helps with to…

    …rquebox.apps not always being available via jmx.
    tobias committed Mar 7, 2011
  2. Expose MessageProcessor status.

    tobias committed Mar 7, 2011
  3. Render static assets correctly for contexts without a slash prefix (T…

    …ORQUE-293)
    
    The existing rails3 asset test w/ a context path was adjusted to
    account for this scenario.
    bbrowning committed Mar 7, 2011
Commits on Mar 5, 2011
  1. Defer any reference to org.infinispan until needed as doing otherwise…

    … breaks the :nothing fallback.
    jcrossley3 committed Mar 5, 2011
Commits on Mar 4, 2011
  1. Now configuring caches according to a particular clustering mode [TOR…

    …QUE-33]
    
    We support the various infinispan modes: replicated, distributed, and
    invalidation, which is the default.  So to have an asynchonous
    replicated cache store:
    
      config.cache_store = :torque_box_store, {:mode => :replicated}
    
    Which is equivalent to this:
    
      config.cache_store = ActiveSupport::Cache::TorqueBoxStore.new(:mode => :replicated)
    
    The :sync option determines whether the write calls wait for acks from
    other members of the cluster:
    
      config.cache_store = :torque_box_store, {:mode => :replicated, :sync => true}
    
    It's also possible to configure multiple TB cache stores in your app,
    each configured to be in a different clustering mode, but the
    developer is responsible for uniquely naming them:
    
      ActiveSupport::Cache::TorqueBoxStore.new(:name => "models", :mode => :replicated)
    
    Docs forthcoming!
    jcrossley3 committed Mar 4, 2011
  2. Add defaults to RackAppMetaData. Default includes config.ru and a con…

    …text.
    
    Rails2 deployments simply erase the rackup for synthetic rackups.
    Rename deployer to lose gratuitous "Default".  Considered adding "Abstract", "Factory" and "Impl", but decided against.
    Removed un-used dead deployer.
    bobmcwhirter committed Mar 4, 2011
Commits on Mar 3, 2011