Permalink
Switch branches/tags
Nothing to show
Commits on Jun 13, 2013
  1. Update README.md

    Somewhat overstated project status.
    committed Jun 13, 2013
Commits on Mar 25, 2010
  1. License notice

    Michael Bridgen committed Mar 25, 2010
Commits on Sep 28, 2009
  1. Supply the username when creating the streams user

    Michael Bridgen committed Sep 28, 2009
  2. Merge backout changeset

    Michael Bridgen committed Sep 28, 2009
  3. Backed out changeset 2faae8040493: restore the Python harness, only to

    have the libraries that are used from the sbin/*.py scripts and the
    plugin tests harness.
    Michael Bridgen committed Sep 28, 2009
Commits on Sep 25, 2009
  1. Bug 21481: Actually I think it is OK, and sensible, to have a reentra…

    …nt lock on publish
    Michael Bridgen committed Sep 25, 2009
  2. Switch debug and trace on for when running the orchestrator from the …

    …Makefile
    Michael Bridgen committed Sep 25, 2009
  3. Make sure we have XML messages for use with with xpathselect, for exa…

    …mple
    Michael Bridgen committed Sep 25, 2009
  4. Chart the trace data, fix up other charts

    Michael Bridgen committed Sep 25, 2009
Commits on Sep 24, 2009
  1. Bug 21481: Test for overlapping synchronous methods on the AMQP

    channel.  It's non-deterministic, because of threads and
    Thread.sleep(); maybe synchronize{} and notify() would help.
    
    This fails with the code from before commit 8d363fa4b798 (provided one
    changes AMQPMessageChannel to AMQPComponentMessageChannel, as it used
    to be), and succeeds now.
    Michael Bridgen committed Sep 24, 2009
  2. Bug 21481 and bug 21512: Don't try to enter the monitor for publish,

    since that will be invoked in handlers, which are already inside.  And
    since servers *won't* usually be publishing in handlers, commit for
    them on publish, otherwise messages will never enter the system.
    
    There's still a race here: publish for servers /ought/ to enter
    the monitor, since publish has to deliver more than one frame and may
    result in interleaving frames.  But this would have the same problem
    as above for the relay server, since it *is* responding to an input
    when it publishes.  So there's a minor danger that either an ack or a
    commit on the command input will overlap with a publish on the
    output.
    Michael Bridgen committed Sep 24, 2009
  3. Bug21481: Have one implementation of MessageChannel, which mutexes all

    synchronous channel operations, as well as only ever using a
    serialising input consumer.  This means we no longer have the problem
    of different things synchronising on different locks, and
    overlapping.
    My worry is that it will introduce deadlocks, or the possibility of
    writing a plugin that deadlocks, because the same mutex object is
    given to the input consumer.  I think this would only be a problem if
    a plugin mixed "phases"; e.g., trying to register a handler while
    handling a message (in fact, yes indeed, that would deadlock).  A
    way to fix that, if we were to do it before reforming the harness
    model, would be to not let the plugin hold on to an interface with
    which it can register handlers.  Anyway.
    Michael Bridgen committed Sep 24, 2009
  4. Have a max size for the sub-process output accumulator, and flush it

    when it gets that big.  This avoids a leak when plugins are behaving
    badly (they are not supposed to output on stdout or stderr).
    Also, report non-zero terminations as errors.
    Michael Bridgen committed Sep 24, 2009
Commits on Sep 23, 2009
  1. Stop tracing model index requests

    Michael Bridgen committed Sep 23, 2009
Commits on Sep 15, 2009
  1. Removing window plugin for the time being (in its current state it's …

    …more
    
    confusing than helpful although it should go in again eventually).
    aschmolck committed Sep 15, 2009
Commits on Sep 14, 2009
  1. Add a streams user/group and other small fiddles with streams.spec

    --HG--
    extra : rebase_source : 383b3fdae6e2c292a152142996861511343dff47
    aschmolck committed Sep 14, 2009
  2. Added python-mako as dependency to Makfile for bin/create_plugin

    --HG--
    extra : rebase_source : 3373b3db602ee7d2f13da52f8f807c2fd89b011f
    aschmolck committed Sep 14, 2009
  3. Added scruffy plugin template generation script and automated plugins…

    …/pom.xml generation.
    
    --HG--
    extra : rebase_source : b927eb1cb3f4a593bb7cd04c0070904293b8b813
    aschmolck committed Sep 14, 2009
  4. Brought getting_started_dev.org somewhat up-to-date wrt. plugins (and…

    … culled stuff).
    
    --HG--
    extra : rebase_source : 54b6f36b0f4c03e5259314ec205492dab24d8144
    aschmolck committed Sep 14, 2009
  5. Merge in hand-tooled webmachine'd API

    --HG--
    files : orchestrator/ebin/orchestrator.app
    files : orchestrator/scripts/streams-server
    files : orchestrator/src/api_root_resource.erl
    files : orchestrator/src/api_web.erl
    files : orchestrator/src/orchestrator.erl
    files : orchestrator/src/orchestrator_root.erl
    files : orchestrator/src/streams_config.erl
    Michael Bridgen committed Sep 14, 2009
Commits on Sep 11, 2009
  1. Create pipelines by POSTing to /model/pipeline

    --HG--
    branch : bug21412
    Michael Bridgen committed Sep 11, 2009
  2. Tests for PUT pipeline and a passing implementation

    --HG--
    branch : bug21412
    Michael Bridgen committed Sep 11, 2009
  3. Use a hitherto not-commonly-used port for the API; allow it, and the …

    …API logs directory, to be supplied in the app env
    
    --HG--
    branch : bug21412
    Michael Bridgen committed Sep 11, 2009
  4. Add uptime to the streamsctl status output, and factor it out into or…

    …chestrator_util. Update usage with debug and trace.
    Michael Bridgen committed Sep 11, 2009
  5. Incomplete implementation of PUT pipeline; needs tests before I can g…

    …o further
    
    --HG--
    branch : bug21412
    Michael Bridgen committed Sep 11, 2009
Commits on Sep 10, 2009
  1. GET a pipeline resource

    --HG--
    branch : bug21412
    Michael Bridgen committed Sep 10, 2009
  2. Whacked python harness

    aschmolck committed Sep 10, 2009
Commits on Sep 9, 2009
  1. A raw network sender, to try sending and receiving without (necessari…

    …ly) streams in the middle
    Michael Bridgen committed Sep 9, 2009
  2. Use the Python stdlib blocking queue rather than our own

    Michael Bridgen committed Sep 9, 2009
Commits on Sep 8, 2009
  1. Merge.

    --HG--
    emptychangelog : true
    aschmolck committed Sep 8, 2009