add timeout for unmet dependencies in task scheduler

IPython/zmq/parallel/asyncresult.py

@@ -1,200 +1,205 b''
 """AsyncResult objects for the client"""
 #-----------------------------------------------------------------------------
 #  Copyright (C) 2010  The IPython Development Team
 #
 #  Distributed under the terms of the BSD License.  The full license is in
 #  the file COPYING, distributed as part of this software.
 #-----------------------------------------------------------------------------
 
 #-----------------------------------------------------------------------------
 # Imports
 #-----------------------------------------------------------------------------
 
 from IPython.external.decorator import decorator
 import error
 
 #-----------------------------------------------------------------------------
 # Classes
 #-----------------------------------------------------------------------------
 
 @decorator
 def check_ready(f, self, *args, **kwargs):
     """Call spin() to sync state prior to calling the method."""
     self.wait(0)
     if not self._ready:
         raise error.TimeoutError("result not ready")
     return f(self, *args, **kwargs)
 
 class AsyncResult(object):
     """Class for representing results of non-blocking calls.
     
     Provides the same interface as :py:class:`multiprocessing.AsyncResult`.
     """
     def __init__(self, client, msg_ids, fname=''):
         self._client = client
         self.msg_ids = msg_ids
         self._fname=fname
         self._ready = False
         self._success = None
-        self._flatten_result = len(msg_ids) == 1
+        self._single_result = len(msg_ids) == 1
     
     def __repr__(self):
         if self._ready:
             return "<%s: finished>"%(self.__class__.__name__)
         else:
             return "<%s: %s>"%(self.__class__.__name__,self._fname)
     
     
     def _reconstruct_result(self, res):
         """
         Override me in subclasses for turning a list of results
         into the expected form.
         """
-        if self._flatten_result:
+        if self._single_result:
             return res[0]
         else:
             return res
         
     def get(self, timeout=-1):
         """Return the result when it arrives. 
         
         If `timeout` is not ``None`` and the result does not arrive within
         `timeout` seconds then ``TimeoutError`` is raised. If the
         remote call raised an exception then that exception will be reraised
         by get().
         """
         if not self.ready():
             self.wait(timeout)
         
         if self._ready:
             if self._success:
                 return self._result
             else:
                 raise self._exception
         else:
             raise error.TimeoutError("Result not ready.")
     
     def ready(self):
         """Return whether the call has completed."""
         if not self._ready:
             self.wait(0)
         return self._ready
     
     def wait(self, timeout=-1):
         """Wait until the result is available or until `timeout` seconds pass.
         """
         if self._ready:
             return
         self._ready = self._client.barrier(self.msg_ids, timeout)
         if self._ready:
             try:
                 results = map(self._client.results.get, self.msg_ids)
                 self._result = results
-                results = error.collect_exceptions(results, self._fname)
+                if self._single_result:
+                    r = results[0]
+                    if isinstance(r, Exception):
+                        raise r
+                else:
+                    results = error.collect_exceptions(results, self._fname)
                 self._result = self._reconstruct_result(results)
             except Exception, e:
                 self._exception = e
                 self._success = False
             else:
                 self._success = True
             finally:
                 self._metadata = map(self._client.metadata.get, self.msg_ids)
             
     
     def successful(self):
         """Return whether the call completed without raising an exception. 
         
         Will raise ``AssertionError`` if the result is not ready.
         """
         assert self._ready
         return self._success
     
     #----------------------------------------------------------------
     # Extra methods not in mp.pool.AsyncResult
     #----------------------------------------------------------------
     
     def get_dict(self, timeout=-1):
         """Get the results as a dict, keyed by engine_id."""
         results = self.get(timeout)
         engine_ids = [ md['engine_id'] for md in self._metadata ]
         bycount = sorted(engine_ids, key=lambda k: engine_ids.count(k))
         maxcount = bycount.count(bycount[-1])
         if maxcount > 1:
             raise ValueError("Cannot build dict, %i jobs ran on engine #%i"%(
                     maxcount, bycount[-1]))
         
         return dict(zip(engine_ids,results))
     
     @property
     @check_ready
     def result(self):
         """result property."""
         return self._result
     
     # abbreviated alias:
     r = result
     
     @property
     @check_ready
     def metadata(self):
         """metadata property."""
-        if self._flatten_result:
+        if self._single_result:
             return self._metadata[0]
         else:
             return self._metadata
     
     @property
     def result_dict(self):
         """result property as a dict."""
         return self.get_dict(0)
     
     def __dict__(self):
         return self.get_dict(0)
 
     #-------------------------------------
     # dict-access
     #-------------------------------------
     
     @check_ready
     def __getitem__(self, key):
         """getitem returns result value(s) if keyed by int/slice, or metadata if key is str.
         """
         if isinstance(key, int):
             return error.collect_exceptions([self._result[key]], self._fname)[0]
         elif isinstance(key, slice):
             return error.collect_exceptions(self._result[key], self._fname)
         elif isinstance(key, basestring):
             values = [ md[key] for md in self._metadata ]
-            if self._flatten_result:
+            if self._single_result:
                 return values[0]
             else:
                 return values
         else:
             raise TypeError("Invalid key type %r, must be 'int','slice', or 'str'"%type(key))
     
     @check_ready
     def __getattr__(self, key):
         """getattr maps to getitem for convenient access to metadata."""
         if key not in self._metadata[0].keys():
             raise AttributeError("%r object has no attribute %r"%(
                     self.__class__.__name__, key))
         return self.__getitem__(key)
 
     
 class AsyncMapResult(AsyncResult):
     """Class for representing results of non-blocking gathers.
     
     This will properly reconstruct the gather.
     """
     
     def __init__(self, client, msg_ids, mapObject, fname=''):
         AsyncResult.__init__(self, client, msg_ids, fname=fname)
         self._mapObject = mapObject
-        self._flatten_result = False
+        self._single_result = False
     
     def _reconstruct_result(self, res):
         """Perform the gather on the actual results."""
         return self._mapObject.joinPartitions(res)
     
         
 __all__ = ['AsyncResult', 'AsyncMapResult']
  No newline at end of file

IPython/zmq/parallel/client.py

@@ -1,1206 +1,1216 b''
 """A semi-synchronous Client for the ZMQ controller"""
 #-----------------------------------------------------------------------------
 #  Copyright (C) 2010  The IPython Development Team
 #
 #  Distributed under the terms of the BSD License.  The full license is in
 #  the file COPYING, distributed as part of this software.
 #-----------------------------------------------------------------------------
 
 #-----------------------------------------------------------------------------
 # Imports
 #-----------------------------------------------------------------------------
 
 import os
 import time
 from getpass import getpass
 from pprint import pprint
 from datetime import datetime
 
 import zmq
 from zmq.eventloop import ioloop, zmqstream
 
 from IPython.external.decorator import decorator
 from IPython.zmq import tunnel
 
 import streamsession as ss
 # from remotenamespace import RemoteNamespace
 from view import DirectView, LoadBalancedView
 from dependency import Dependency, depend, require
 import error
 import map as Map
 from asyncresult import AsyncResult, AsyncMapResult
 from remotefunction import remote,parallel,ParallelFunction,RemoteFunction
 from util import ReverseDict
 
 #--------------------------------------------------------------------------
 # helpers for implementing old MEC API via client.apply
 #--------------------------------------------------------------------------
 
 def _push(ns):
     """helper method for implementing `client.push` via `client.apply`"""
     globals().update(ns)
 
 def _pull(keys):
     """helper method for implementing `client.pull` via `client.apply`"""
     g = globals()
     if isinstance(keys, (list,tuple, set)):
         for key in keys:
             if not g.has_key(key):
                 raise NameError("name '%s' is not defined"%key)
         return map(g.get, keys)
     else:
         if not g.has_key(keys):
             raise NameError("name '%s' is not defined"%keys)
         return g.get(keys)
 
 def _clear():
     """helper method for implementing `client.clear` via `client.apply`"""
     globals().clear()
 
 def _execute(code):
     """helper method for implementing `client.execute` via `client.apply`"""
     exec code in globals()
     
 
 #--------------------------------------------------------------------------
 # Decorators for Client methods
 #--------------------------------------------------------------------------
 
 @decorator
 def spinfirst(f, self, *args, **kwargs):
     """Call spin() to sync state prior to calling the method."""
     self.spin()
     return f(self, *args, **kwargs)
 
 @decorator
 def defaultblock(f, self, *args, **kwargs):
     """Default to self.block; preserve self.block."""
     block = kwargs.get('block',None)
     block = self.block if block is None else block
     saveblock = self.block
     self.block = block
     try:
         ret = f(self, *args, **kwargs)
     finally:
         self.block = saveblock
     return ret
 
 
 #--------------------------------------------------------------------------
 # Classes
 #--------------------------------------------------------------------------
 
 class Metadata(dict):
     """Subclass of dict for initializing metadata values.
     
     Attribute access works on keys.
     
     These objects have a strict set of keys - errors will raise if you try
     to add new keys.
     """
     def __init__(self, *args, **kwargs):
         dict.__init__(self)
         md = {'msg_id' : None,
               'submitted' : None,
               'started' : None,
               'completed' : None,
               'received' : None,
               'engine_uuid' : None,
               'engine_id' : None,
               'follow' : None,
               'after' : None,
               'status' : None,
 
               'pyin' : None,
               'pyout' : None,
               'pyerr' : None,
               'stdout' : '',
               'stderr' : '',
             }
         self.update(md)
         self.update(dict(*args, **kwargs))
     
     def __getattr__(self, key):
         """getattr aliased to getitem"""
         if key in self.iterkeys():
             return self[key]
         else:
             raise AttributeError(key)
 
     def __setattr__(self, key, value):
         """setattr aliased to setitem, with strict"""
         if key in self.iterkeys():
             self[key] = value
         else:
             raise AttributeError(key)
     
     def __setitem__(self, key, value):
         """strict static key enforcement"""
         if key in self.iterkeys():
             dict.__setitem__(self, key, value)
         else:
             raise KeyError(key)
         
 
 class Client(object):
     """A semi-synchronous client to the IPython ZMQ controller
     
     Parameters
     ----------
     
     addr : bytes; zmq url, e.g. 'tcp://127.0.0.1:10101'
         The address of the controller's registration socket.
         [Default: 'tcp://127.0.0.1:10101']
     context : zmq.Context
         Pass an existing zmq.Context instance, otherwise the client will create its own
     username : bytes
         set username to be passed to the Session object
     debug : bool
         flag for lots of message printing for debug purposes
  
     #-------------- ssh related args ----------------
     # These are args for configuring the ssh tunnel to be used
     # credentials are used to forward connections over ssh to the Controller
     # Note that the ip given in `addr` needs to be relative to sshserver
     # The most basic case is to leave addr as pointing to localhost (127.0.0.1),
     # and set sshserver as the same machine the Controller is on. However, 
     # the only requirement is that sshserver is able to see the Controller
     # (i.e. is within the same trusted network).
     
     sshserver : str
         A string of the form passed to ssh, i.e. 'server.tld' or '[e-mail]:port'
         If keyfile or password is specified, and this is not, it will default to
         the ip given in addr.
     sshkey : str; path to public ssh key file
         This specifies a key to be used in ssh login, default None.
         Regular default ssh keys will be used without specifying this argument.
     password : str; 
         Your ssh password to sshserver. Note that if this is left None,
         you will be prompted for it if passwordless key based login is unavailable.
     
     #------- exec authentication args -------
     # If even localhost is untrusted, you can have some protection against
     # unauthorized execution by using a key.  Messages are still sent
     # as cleartext, so if someone can snoop your loopback traffic this will
     # not help anything.
     
     exec_key : str
         an authentication key or file containing a key
         default: None
     
     
     Attributes
     ----------
     ids : set of int engine IDs
         requesting the ids attribute always synchronizes
         the registration state. To request ids without synchronization,
         use semi-private _ids attributes.
     
     history : list of msg_ids
         a list of msg_ids, keeping track of all the execution
         messages you have submitted in order.
     
     outstanding : set of msg_ids
         a set of msg_ids that have been submitted, but whose
         results have not yet been received.
     
     results : dict
         a dict of all our results, keyed by msg_id
     
     block : bool
         determines default behavior when block not specified
         in execution methods
     
     Methods
     -------
     spin : flushes incoming results and registration state changes
             control methods spin, and requesting `ids` also ensures up to date
             
     barrier : wait on one or more msg_ids
     
     execution methods: apply/apply_bound/apply_to/apply_bound
         legacy: execute, run
     
     query methods: queue_status, get_result, purge
     
     control methods: abort, kill
     
     """
     
     
     _connected=False
     _ssh=False
     _engines=None
     _addr='tcp://127.0.0.1:10101'
     _registration_socket=None
     _query_socket=None
     _control_socket=None
     _iopub_socket=None
     _notification_socket=None
     _mux_socket=None
     _task_socket=None
     block = False
     outstanding=None
     results = None
     history = None
     debug = False
     targets = None
     
     def __init__(self, addr='tcp://127.0.0.1:10101', context=None, username=None, debug=False, 
             sshserver=None, sshkey=None, password=None, paramiko=None,
             exec_key=None,):
         if context is None:
             context = zmq.Context()
         self.context = context
         self.targets = 'all'
         self._addr = addr
         self._ssh = bool(sshserver or sshkey or password)
         if self._ssh and sshserver is None:
             # default to the same
             sshserver = addr.split('://')[1].split(':')[0]
         if self._ssh and password is None:
             if tunnel.try_passwordless_ssh(sshserver, sshkey, paramiko):
                 password=False
             else:
                 password = getpass("SSH Password for %s: "%sshserver)
         ssh_kwargs = dict(keyfile=sshkey, password=password, paramiko=paramiko)
         
         if exec_key is not None and os.path.isfile(exec_key):
             arg = 'keyfile'
         else:
             arg = 'key'
         key_arg = {arg:exec_key}
         if username is None:
             self.session = ss.StreamSession(**key_arg)
         else:
             self.session = ss.StreamSession(username, **key_arg)
         self._registration_socket = self.context.socket(zmq.XREQ)
         self._registration_socket.setsockopt(zmq.IDENTITY, self.session.session)
         if self._ssh:
             tunnel.tunnel_connection(self._registration_socket, addr, sshserver, **ssh_kwargs)
         else:
             self._registration_socket.connect(addr)
         self._engines = ReverseDict()
         self._ids = set()
         self.outstanding=set()
         self.results = {}
         self.metadata = {}
         self.history = []
         self.debug = debug
         self.session.debug = debug
         
         self._notification_handlers = {'registration_notification' : self._register_engine,
                                     'unregistration_notification' : self._unregister_engine,
                                     }
         self._queue_handlers = {'execute_reply' : self._handle_execute_reply,
                                 'apply_reply' : self._handle_apply_reply}
         self._connect(sshserver, ssh_kwargs)
         
     
     @property
     def ids(self):
         """Always up to date ids property."""
         self._flush_notifications()
         return self._ids
     
     def _update_engines(self, engines):
         """Update our engines dict and _ids from a dict of the form: {id:uuid}."""
         for k,v in engines.iteritems():
             eid = int(k)
             self._engines[eid] = bytes(v) # force not unicode
             self._ids.add(eid)
     
     def _build_targets(self, targets):
         """Turn valid target IDs or 'all' into two lists:
         (int_ids, uuids).
         """
         if targets is None:
             targets = self._ids
         elif isinstance(targets, str):
             if targets.lower() == 'all':
                 targets = self._ids
             else:
                 raise TypeError("%r not valid str target, must be 'all'"%(targets))
         elif isinstance(targets, int):
             targets = [targets]
         return [self._engines[t] for t in targets], list(targets)
     
     def _connect(self, sshserver, ssh_kwargs):
         """setup all our socket connections to the controller. This is called from
         __init__."""
         if self._connected:
             return
         self._connected=True
 
         def connect_socket(s, addr):
             if self._ssh:
                 return tunnel.tunnel_connection(s, addr, sshserver, **ssh_kwargs)
             else:
                 return s.connect(addr)
             
         self.session.send(self._registration_socket, 'connection_request')
         idents,msg = self.session.recv(self._registration_socket,mode=0)
         if self.debug:
             pprint(msg)
         msg = ss.Message(msg)
         content = msg.content
         if content.status == 'ok':
             if content.mux:
                 self._mux_socket = self.context.socket(zmq.PAIR)
                 self._mux_socket.setsockopt(zmq.IDENTITY, self.session.session)
                 connect_socket(self._mux_socket, content.mux)
             if content.task:
                 self._task_socket = self.context.socket(zmq.PAIR)
                 self._task_socket.setsockopt(zmq.IDENTITY, self.session.session)
                 connect_socket(self._task_socket, content.task)
             if content.notification:
                 self._notification_socket = self.context.socket(zmq.SUB)
                 connect_socket(self._notification_socket, content.notification)
                 self._notification_socket.setsockopt(zmq.SUBSCRIBE, "")
             if content.query:
                 self._query_socket = self.context.socket(zmq.PAIR)
                 self._query_socket.setsockopt(zmq.IDENTITY, self.session.session)
                 connect_socket(self._query_socket, content.query)
             if content.control:
                 self._control_socket = self.context.socket(zmq.PAIR)
                 self._control_socket.setsockopt(zmq.IDENTITY, self.session.session)
                 connect_socket(self._control_socket, content.control)
             if content.iopub:
                 self._iopub_socket = self.context.socket(zmq.SUB)
                 self._iopub_socket.setsockopt(zmq.SUBSCRIBE, '')
                 self._iopub_socket.setsockopt(zmq.IDENTITY, self.session.session)
                 connect_socket(self._iopub_socket, content.iopub)
             self._update_engines(dict(content.engines))
                 
         else:
             self._connected = False
             raise Exception("Failed to connect!")
     
     #--------------------------------------------------------------------------
     # handlers and callbacks for incoming messages
     #--------------------------------------------------------------------------
     
     def _register_engine(self, msg):
         """Register a new engine, and update our connection info."""
         content = msg['content']
         eid = content['id']
         d = {eid : content['queue']}
         self._update_engines(d)
         self._ids.add(int(eid))
 
     def _unregister_engine(self, msg):
         """Unregister an engine that has died."""
         content = msg['content']
         eid = int(content['id'])
         if eid in self._ids:
             self._ids.remove(eid)
             self._engines.pop(eid)
     
     def _extract_metadata(self, header, parent, content):
         md = {'msg_id' : parent['msg_id'],
               'received' : datetime.now(),
               'engine_uuid' : header.get('engine', None),
               'follow' : parent['follow'],
               'after' : parent['after'],
               'status' : content['status'],
             }
         
         if md['engine_uuid'] is not None:
             md['engine_id'] = self._engines.get(md['engine_uuid'], None)
         
         if 'date' in parent:
             md['submitted'] = datetime.strptime(parent['date'], ss.ISO8601)
         if 'started' in header:
             md['started'] = datetime.strptime(header['started'], ss.ISO8601)
         if 'date' in header:
             md['completed'] = datetime.strptime(header['date'], ss.ISO8601)
         return md
     
     def _handle_execute_reply(self, msg):
         """Save the reply to an execute_request into our results.
         
         execute messages are never actually used. apply is used instead.
         """
         
         parent = msg['parent_header']
         msg_id = parent['msg_id']
         if msg_id not in self.outstanding:
             if msg_id in self.history:
                 print ("got stale result: %s"%msg_id)
             else:
                 print ("got unknown result: %s"%msg_id)
         else:
             self.outstanding.remove(msg_id)
         self.results[msg_id] = ss.unwrap_exception(msg['content'])
     
     def _handle_apply_reply(self, msg):
         """Save the reply to an apply_request into our results."""
         parent = msg['parent_header']
         msg_id = parent['msg_id']
         if msg_id not in self.outstanding:
             if msg_id in self.history:
                 print ("got stale result: %s"%msg_id)
                 print self.results[msg_id]
                 print msg
             else:
                 print ("got unknown result: %s"%msg_id)
         else:
             self.outstanding.remove(msg_id)
         content = msg['content']
         header = msg['header']
         
         # construct metadata:
         md = self.metadata.setdefault(msg_id, Metadata())
         md.update(self._extract_metadata(header, parent, content))
         self.metadata[msg_id] = md
         
         # construct result:
         if content['status'] == 'ok':
             self.results[msg_id] = ss.unserialize_object(msg['buffers'])[0]
         elif content['status'] == 'aborted':
             self.results[msg_id] = error.AbortedTask(msg_id)
         elif content['status'] == 'resubmitted':
             # TODO: handle resubmission
             pass
         else:
             e = ss.unwrap_exception(content)
             if e.engine_info:
                 e_uuid = e.engine_info['engineid']
                 eid = self._engines[e_uuid]
                 e.engine_info['engineid'] = eid
             self.results[msg_id] = e
     
     def _flush_notifications(self):
         """Flush notifications of engine registrations waiting
         in ZMQ queue."""
         msg = self.session.recv(self._notification_socket, mode=zmq.NOBLOCK)
         while msg is not None:
             if self.debug:
                 pprint(msg)
             msg = msg[-1]
             msg_type = msg['msg_type']
             handler = self._notification_handlers.get(msg_type, None)
             if handler is None:
                 raise Exception("Unhandled message type: %s"%msg.msg_type)
             else:
                 handler(msg)
             msg = self.session.recv(self._notification_socket, mode=zmq.NOBLOCK)
     
     def _flush_results(self, sock):
         """Flush task or queue results waiting in ZMQ queue."""
         msg = self.session.recv(sock, mode=zmq.NOBLOCK)
         while msg is not None:
             if self.debug:
                 pprint(msg)
             msg = msg[-1]
             msg_type = msg['msg_type']
             handler = self._queue_handlers.get(msg_type, None)
             if handler is None:
                 raise Exception("Unhandled message type: %s"%msg.msg_type)
             else:
                 handler(msg)
             msg = self.session.recv(sock, mode=zmq.NOBLOCK)
     
     def _flush_control(self, sock):
         """Flush replies from the control channel waiting
         in the ZMQ queue.
         
         Currently: ignore them."""
         msg = self.session.recv(sock, mode=zmq.NOBLOCK)
         while msg is not None:
             if self.debug:
                 pprint(msg)
             msg = self.session.recv(sock, mode=zmq.NOBLOCK)
     
     def _flush_iopub(self, sock):
         """Flush replies from the iopub channel waiting
         in the ZMQ queue.
         """
         msg = self.session.recv(sock, mode=zmq.NOBLOCK)
         while msg is not None:
             if self.debug:
                 pprint(msg)
             msg = msg[-1]
             parent = msg['parent_header']
             msg_id = parent['msg_id']
             content = msg['content']
             header = msg['header']
             msg_type = msg['msg_type']
             
             # init metadata:
             md = self.metadata.setdefault(msg_id, Metadata())
             
             if msg_type == 'stream':
                 name = content['name']
                 s = md[name] or ''
                 md[name] = s + content['data']
             elif msg_type == 'pyerr':
                 md.update({'pyerr' : ss.unwrap_exception(content)})
             else:
                 md.update({msg_type : content['data']})
             
             self.metadata[msg_id] = md
             
             msg = self.session.recv(sock, mode=zmq.NOBLOCK)
     
     #--------------------------------------------------------------------------
     # getitem
     #--------------------------------------------------------------------------
     
     def __getitem__(self, key):
         """Dict access returns DirectView multiplexer objects or,
         if key is None, a LoadBalancedView."""
         if key is None:
             return LoadBalancedView(self)
         if isinstance(key, int):
             if key not in self.ids:
                 raise IndexError("No such engine: %i"%key)
             return DirectView(self, key)
         
         if isinstance(key, slice):
             indices = range(len(self.ids))[key]
             ids = sorted(self._ids)
             key = [ ids[i] for i in indices ]
             # newkeys = sorted(self._ids)[thekeys[k]]
         
         if isinstance(key, (tuple, list, xrange)):
             _,targets = self._build_targets(list(key))
             return DirectView(self, targets)
         else:
             raise TypeError("key by int/iterable of ints only, not %s"%(type(key)))
     
     #--------------------------------------------------------------------------
     # Begin public methods
     #--------------------------------------------------------------------------
     
     @property
     def remote(self):
         """property for convenient RemoteFunction generation.
         
         >>> @client.remote
         ... def f():
                 import os
                 print (os.getpid())
         """
         return remote(self, block=self.block)
     
     def spin(self):
         """Flush any registration notifications and execution results
         waiting in the ZMQ queue.
         """
         if self._notification_socket:
             self._flush_notifications()
         if self._mux_socket:
             self._flush_results(self._mux_socket)
         if self._task_socket:
             self._flush_results(self._task_socket)
         if self._control_socket:
             self._flush_control(self._control_socket)
         if self._iopub_socket:
             self._flush_iopub(self._iopub_socket)
     
     def barrier(self, msg_ids=None, timeout=-1):
         """waits on one or more `msg_ids`, for up to `timeout` seconds.
         
         Parameters
         ----------
         msg_ids : int, str, or list of ints and/or strs, or one or more AsyncResult objects
                 ints are indices to self.history
                 strs are msg_ids
                 default: wait on all outstanding messages
         timeout : float
                 a time in seconds, after which to give up.
                 default is -1, which means no timeout
         
         Returns
         -------
         True : when all msg_ids are done
         False : timeout reached, some msg_ids still outstanding
         """
         tic = time.time()
         if msg_ids is None:
             theids = self.outstanding
         else:
             if isinstance(msg_ids, (int, str, AsyncResult)):
                 msg_ids = [msg_ids]
             theids = set()
             for msg_id in msg_ids:
                 if isinstance(msg_id, int):
                     msg_id = self.history[msg_id]
                 elif isinstance(msg_id, AsyncResult):
                     map(theids.add, msg_id.msg_ids)
                     continue
                 theids.add(msg_id)
         if not theids.intersection(self.outstanding):
             return True
         self.spin()
         while theids.intersection(self.outstanding):
             if timeout >= 0 and ( time.time()-tic ) > timeout:
                 break
             time.sleep(1e-3)
             self.spin()
         return len(theids.intersection(self.outstanding)) == 0
     
     #--------------------------------------------------------------------------
     # Control methods
     #--------------------------------------------------------------------------
     
     @spinfirst
     @defaultblock
     def clear(self, targets=None, block=None):
         """Clear the namespace in target(s)."""
         targets = self._build_targets(targets)[0]
         for t in targets:
             self.session.send(self._control_socket, 'clear_request', content={}, ident=t)
         error = False
         if self.block:
             for i in range(len(targets)):
                 idents,msg = self.session.recv(self._control_socket,0)
                 if self.debug:
                     pprint(msg)
                 if msg['content']['status'] != 'ok':
                     error = ss.unwrap_exception(msg['content'])
         if error:
             return error
         
     
     @spinfirst
     @defaultblock
     def abort(self, msg_ids = None, targets=None, block=None):
         """Abort the execution queues of target(s)."""
         targets = self._build_targets(targets)[0]
         if isinstance(msg_ids, basestring):
             msg_ids = [msg_ids]
         content = dict(msg_ids=msg_ids)
         for t in targets:
             self.session.send(self._control_socket, 'abort_request', 
                     content=content, ident=t)
         error = False
         if self.block:
             for i in range(len(targets)):
                 idents,msg = self.session.recv(self._control_socket,0)
                 if self.debug:
                     pprint(msg)
                 if msg['content']['status'] != 'ok':
                     error = ss.unwrap_exception(msg['content'])
         if error:
             return error
     
     @spinfirst
     @defaultblock
     def shutdown(self, targets=None, restart=False, controller=False, block=None):
         """Terminates one or more engine processes, optionally including the controller."""
         if controller:
             targets = 'all'
         targets = self._build_targets(targets)[0]
         for t in targets:
             self.session.send(self._control_socket, 'shutdown_request', 
                         content={'restart':restart},ident=t)
         error = False
         if block or controller:
             for i in range(len(targets)):
                 idents,msg = self.session.recv(self._control_socket,0)
                 if self.debug:
                     pprint(msg)
                 if msg['content']['status'] != 'ok':
                     error = ss.unwrap_exception(msg['content'])
         
         if controller:
             time.sleep(0.25)
             self.session.send(self._query_socket, 'shutdown_request')
             idents,msg = self.session.recv(self._query_socket, 0)
             if self.debug:
                 pprint(msg)
             if msg['content']['status'] != 'ok':
                 error = ss.unwrap_exception(msg['content'])
         
         if error:
             raise error
     
     #--------------------------------------------------------------------------
     # Execution methods
     #--------------------------------------------------------------------------
     
     @defaultblock
     def execute(self, code, targets='all', block=None):
         """Executes `code` on `targets` in blocking or nonblocking manner.
         
         ``execute`` is always `bound` (affects engine namespace)
         
         Parameters
         ----------
         code : str
                 the code string to be executed
         targets : int/str/list of ints/strs
                 the engines on which to execute
                 default : all
         block : bool
                 whether or not to wait until done to return
                 default: self.block
         """
         result = self.apply(_execute, (code,), targets=targets, block=self.block, bound=True)
         return result
     
     def run(self, code, block=None):
         """Runs `code` on an engine. 
         
         Calls to this are load-balanced.
         
         ``run`` is never `bound` (no effect on engine namespace)
         
         Parameters
         ----------
         code : str
                 the code string to be executed
         block : bool
                 whether or not to wait until done
         
         """
         result = self.apply(_execute, (code,), targets=None, block=block, bound=False)
         return result
     
     def _maybe_raise(self, result):
         """wrapper for maybe raising an exception if apply failed."""
         if isinstance(result, error.RemoteError):
             raise result
         
         return result
-            
+    
+    def _build_dependency(self, dep):
+        """helper for building jsonable dependencies from various input forms"""
+        if isinstance(dep, Dependency):
+            return dep.as_dict()
+        elif isinstance(dep, AsyncResult):
+            return dep.msg_ids
+        elif dep is None:
+            return []
+        elif isinstance(dep, set):
+            return list(dep)
+        elif isinstance(dep, (list,dict)):
+            return dep
+        elif isinstance(dep, str):
+            return [dep]
+        else:
+            raise TypeError("Dependency may be: set,list,dict,Dependency or AsyncResult, not %r"%type(dep))
+        
     def apply(self, f, args=None, kwargs=None, bound=True, block=None, targets=None,
-                        after=None, follow=None):
+                        after=None, follow=None, timeout=None):
         """Call `f(*args, **kwargs)` on a remote engine(s), returning the result.
         
         This is the central execution command for the client.
         
         Parameters
         ----------
         
         f : function
             The fuction to be called remotely
         args : tuple/list
             The positional arguments passed to `f`
         kwargs : dict
             The keyword arguments passed to `f`
         bound : bool (default: True)
             Whether to execute in the Engine(s) namespace, or in a clean
             namespace not affecting the engine.
         block : bool (default: self.block)
             Whether to wait for the result, or return immediately.
             False:
                 returns msg_id(s)
                 if multiple targets:
                     list of ids
             True:
                 returns actual result(s) of f(*args, **kwargs)
                 if multiple targets:
                     dict of results, by engine ID
         targets : int,list of ints, 'all', None
             Specify the destination of the job.
             if None:
                 Submit via Task queue for load-balancing.
             if 'all':
                 Run on all active engines
             if list:
                 Run on each specified engine
             if int:
                 Run on single engine
             
         after : Dependency or collection of msg_ids
             Only for load-balanced execution (targets=None)
             Specify a list of msg_ids as a time-based dependency.
             This job will only be run *after* the dependencies
             have been met.
             
         follow : Dependency or collection of msg_ids
             Only for load-balanced execution (targets=None)
             Specify a list of msg_ids as a location-based dependency.
             This job will only be run on an engine where this dependency
             is met.
         
+        timeout : float or None
+            Only for load-balanced execution (targets=None)
+            Specify an amount of time (in seconds)
+        
         Returns
         -------
         if block is False:
             if single target:
                 return msg_id
             else:
                 return list of msg_ids
                 ? (should this be dict like block=True) ?
         else:
             if single target:
                 return result of f(*args, **kwargs)
             else:
                 return dict of results, keyed by engine
         """
         
         # defaults:
         block = block if block is not None else self.block
         args = args if args is not None else []
         kwargs = kwargs if kwargs is not None else {}
         
         # enforce types of f,args,kwrags
         if not callable(f):
             raise TypeError("f must be callable, not %s"%type(f))
         if not isinstance(args, (tuple, list)):
             raise TypeError("args must be tuple or list, not %s"%type(args))
         if not isinstance(kwargs, dict):
             raise TypeError("kwargs must be dict, not %s"%type(kwargs))
-            
-        if isinstance(after, Dependency):
-            after = after.as_dict()
-        elif isinstance(after, AsyncResult):
-            after=after.msg_ids
-        elif after is None:
-            after = []
-        if isinstance(follow, Dependency):
-            # if len(follow) > 1 and follow.mode == 'all':
-            #     warn("complex follow-dependencies are not rigorously tested for reachability", UserWarning)
-            follow = follow.as_dict()
-        elif isinstance(follow, AsyncResult):
-            follow=follow.msg_ids
-        elif follow is None:
-            follow = []
-        options  = dict(bound=bound, block=block, after=after, follow=follow)
+        
+        after = self._build_dependency(after)
+        follow = self._build_dependency(follow)
+        
+        options  = dict(bound=bound, block=block)
             
         if targets is None:
-            return self._apply_balanced(f, args, kwargs, **options)
+            return self._apply_balanced(f, args, kwargs, timeout=timeout, 
+                                            after=after, follow=follow, **options)
         else:
             return self._apply_direct(f, args, kwargs, targets=targets, **options)
     
     def _apply_balanced(self, f, args, kwargs, bound=True, block=None,
-                            after=None, follow=None):
+                            after=None, follow=None, timeout=None):
         """The underlying method for applying functions in a load balanced
         manner, via the task queue."""
-        subheader = dict(after=after, follow=follow)
+        subheader = dict(after=after, follow=follow, timeout=timeout)
         bufs = ss.pack_apply_message(f,args,kwargs)
         content = dict(bound=bound)
         
         msg = self.session.send(self._task_socket, "apply_request", 
                 content=content, buffers=bufs, subheader=subheader)
         msg_id = msg['msg_id']
         self.outstanding.add(msg_id)
         self.history.append(msg_id)
         ar = AsyncResult(self, [msg_id], fname=f.__name__)
         if block:
             return ar.get()
         else:
             return ar
     
-    def _apply_direct(self, f, args, kwargs, bound=True, block=None, targets=None,
-                                after=None, follow=None):
+    def _apply_direct(self, f, args, kwargs, bound=True, block=None, targets=None):
         """Then underlying method for applying functions to specific engines
         via the MUX queue."""
         
         queues,targets = self._build_targets(targets)
         
         subheader = dict(after=after, follow=follow)
         content = dict(bound=bound)
         bufs = ss.pack_apply_message(f,args,kwargs)
         
         msg_ids = []
         for queue in queues:
             msg = self.session.send(self._mux_socket, "apply_request", 
                     content=content, buffers=bufs,ident=queue, subheader=subheader)
             msg_id = msg['msg_id']
             self.outstanding.add(msg_id)
             self.history.append(msg_id)
             msg_ids.append(msg_id)
         ar = AsyncResult(self, msg_ids, fname=f.__name__)
         if block:
             return ar.get()
         else:
             return ar
     
     #--------------------------------------------------------------------------
     # Map and decorators
     #--------------------------------------------------------------------------
     
     def map(self, f, *sequences):
         """Parallel version of builtin `map`, using all our engines."""
         pf = ParallelFunction(self, f, block=self.block,
                         bound=True, targets='all')
         return pf.map(*sequences)
     
     def parallel(self, bound=True, targets='all', block=True):
         """Decorator for making a ParallelFunction."""
         return parallel(self, bound=bound, targets=targets, block=block)
     
     def remote(self, bound=True, targets='all', block=True):
         """Decorator for making a RemoteFunction."""
         return remote(self, bound=bound, targets=targets, block=block)
     
     #--------------------------------------------------------------------------
     # Data movement
     #--------------------------------------------------------------------------
     
     @defaultblock
     def push(self, ns, targets='all', block=None):
         """Push the contents of `ns` into the namespace on `target`"""
         if not isinstance(ns, dict):
             raise TypeError("Must be a dict, not %s"%type(ns))
         result = self.apply(_push, (ns,), targets=targets, block=block, bound=True)
         return result
     
     @defaultblock
     def pull(self, keys, targets='all', block=None):
         """Pull objects from `target`'s namespace by `keys`"""
         if isinstance(keys, str):
             pass
         elif isinstance(keys, (list,tuple,set)):
             for key in keys:
                 if not isinstance(key, str):
                     raise TypeError
         result = self.apply(_pull, (keys,), targets=targets, block=block, bound=True)
         return result
     
     def scatter(self, key, seq, dist='b', flatten=False, targets='all', block=None):
         """
         Partition a Python sequence and send the partitions to a set of engines.
         """
         block = block if block is not None else self.block
         targets = self._build_targets(targets)[-1]
         mapObject = Map.dists[dist]()
         nparts = len(targets)
         msg_ids = []
         for index, engineid in enumerate(targets):
             partition = mapObject.getPartition(seq, index, nparts)
             if flatten and len(partition) == 1:
                 r = self.push({key: partition[0]}, targets=engineid, block=False)
             else:
                 r = self.push({key: partition}, targets=engineid, block=False)
             msg_ids.extend(r.msg_ids)
         r = AsyncResult(self, msg_ids, fname='scatter')
         if block:
             return r.get()
         else:
             return r
     
     def gather(self, key, dist='b', targets='all', block=None):
         """
         Gather a partitioned sequence on a set of engines as a single local seq.
         """
         block = block if block is not None else self.block
         
         targets = self._build_targets(targets)[-1]
         mapObject = Map.dists[dist]()
         msg_ids = []
         for index, engineid in enumerate(targets):
             msg_ids.extend(self.pull(key, targets=engineid,block=False).msg_ids)
         
         r = AsyncMapResult(self, msg_ids, mapObject, fname='gather')
         if block:
             return r.get()
         else:
             return r
     
     #--------------------------------------------------------------------------
     # Query methods
     #--------------------------------------------------------------------------
     
     @spinfirst
     def get_results(self, msg_ids, status_only=False):
         """Returns the result of the execute or task request with `msg_ids`.
         
         Parameters
         ----------
         msg_ids : list of ints or msg_ids
             if int:
                 Passed as index to self.history for convenience.
         status_only : bool (default: False)
             if False:
                 return the actual results
         
         Returns
         -------
         
         results : dict
             There will always be the keys 'pending' and 'completed', which will
             be lists of msg_ids.
         """
         if not isinstance(msg_ids, (list,tuple)):
             msg_ids = [msg_ids]
         theids = []
         for msg_id in msg_ids:
             if isinstance(msg_id, int):
                 msg_id = self.history[msg_id]
             if not isinstance(msg_id, str):
                 raise TypeError("msg_ids must be str, not %r"%msg_id)
             theids.append(msg_id)
         
         completed = []
         local_results = {}
         # temporarily disable local shortcut
         # for msg_id in list(theids):
         #     if msg_id in self.results:
         #         completed.append(msg_id)
         #         local_results[msg_id] = self.results[msg_id]
         #         theids.remove(msg_id)
         
         if theids: # some not locally cached
             content = dict(msg_ids=theids, status_only=status_only)
             msg = self.session.send(self._query_socket, "result_request", content=content)
             zmq.select([self._query_socket], [], [])
             idents,msg = self.session.recv(self._query_socket, zmq.NOBLOCK)
             if self.debug:
                 pprint(msg)
             content = msg['content']
             if content['status'] != 'ok':
                 raise ss.unwrap_exception(content)
             buffers = msg['buffers']
         else:
             content = dict(completed=[],pending=[])
         
         content['completed'].extend(completed)
         
         if status_only:
             return content
         
         failures = []
         # load cached results into result:
         content.update(local_results)
         # update cache with results:
         for msg_id in sorted(theids):
             if msg_id in content['completed']:
                 rec = content[msg_id]
                 parent = rec['header']
                 header = rec['result_header']
                 rcontent = rec['result_content']
                 iodict = rec['io']
                 if isinstance(rcontent, str):
                     rcontent = self.session.unpack(rcontent)
                 
                 md = self.metadata.setdefault(msg_id, Metadata())
                 md.update(self._extract_metadata(header, parent, rcontent))
                 md.update(iodict)
                 
                 if rcontent['status'] == 'ok':
                     res,buffers = ss.unserialize_object(buffers)
                 else:
                     res = ss.unwrap_exception(rcontent)
                     failures.append(res)
                 
                 self.results[msg_id] = res
                 content[msg_id] = res
         
         error.collect_exceptions(failures, "get_results")
         return content
 
     @spinfirst
     def queue_status(self, targets=None, verbose=False):
         """Fetch the status of engine queues.
         
         Parameters
         ----------
         targets : int/str/list of ints/strs
                 the engines on which to execute
                 default : all
         verbose : bool
                 Whether to return lengths only, or lists of ids for each element
         """
         targets = self._build_targets(targets)[1]
         content = dict(targets=targets, verbose=verbose)
         self.session.send(self._query_socket, "queue_request", content=content)
         idents,msg = self.session.recv(self._query_socket, 0)
         if self.debug:
             pprint(msg)
         content = msg['content']
         status = content.pop('status')
         if status != 'ok':
             raise ss.unwrap_exception(content)
         return ss.rekey(content)
         
     @spinfirst
     def purge_results(self, msg_ids=[], targets=[]):
         """Tell the controller to forget results.
         
         Individual results can be purged by msg_id, or the entire
         history of specific targets can be purged.
         
         Parameters
         ----------
         msg_ids : str or list of strs
                 the msg_ids whose results should be forgotten.
         targets : int/str/list of ints/strs
                 The targets, by uuid or int_id, whose entire history is to be purged.
                 Use `targets='all'` to scrub everything from the controller's memory.
                 
                 default : None
         """
         if not targets and not msg_ids:
             raise ValueError
         if targets:
             targets = self._build_targets(targets)[1]
         content = dict(targets=targets, msg_ids=msg_ids)
         self.session.send(self._query_socket, "purge_request", content=content)
         idents, msg = self.session.recv(self._query_socket, 0)
         if self.debug:
             pprint(msg)
         content = msg['content']
         if content['status'] != 'ok':
             raise ss.unwrap_exception(content)
 
     #----------------------------------------
     # activate for %px,%autopx magics
     #----------------------------------------
     def activate(self):
         """Make this `View` active for parallel magic commands.
         
         IPython has a magic command syntax to work with `MultiEngineClient` objects.
         In a given IPython session there is a single active one.  While
         there can be many `Views` created and used by the user, 
         there is only one active one.  The active `View` is used whenever 
         the magic commands %px and %autopx are used.
         
         The activate() method is called on a given `View` to make it 
         active.  Once this has been done, the magic commands can be used.
         """
         
         try:
             # This is injected into __builtins__.
             ip = get_ipython()
         except NameError:
             print "The IPython parallel magics (%result, %px, %autopx) only work within IPython."
         else:
             pmagic = ip.plugin_manager.get_plugin('parallelmagic')
             if pmagic is not None:
                 pmagic.active_multiengine_client = self
             else:
                 print "You must first load the parallelmagic extension " \
                       "by doing '%load_ext parallelmagic'"
 
 class AsynClient(Client):
     """An Asynchronous client, using the Tornado Event Loop.
     !!!unfinished!!!"""
     io_loop = None
     _queue_stream = None
     _notifier_stream = None
     _task_stream = None
     _control_stream = None
     
     def __init__(self, addr, context=None, username=None, debug=False, io_loop=None):
         Client.__init__(self, addr, context, username, debug)
         if io_loop is None:
             io_loop = ioloop.IOLoop.instance()
         self.io_loop = io_loop
         
         self._queue_stream = zmqstream.ZMQStream(self._mux_socket, io_loop)
         self._control_stream = zmqstream.ZMQStream(self._control_socket, io_loop)
         self._task_stream = zmqstream.ZMQStream(self._task_socket, io_loop)
         self._notification_stream = zmqstream.ZMQStream(self._notification_socket, io_loop)
     
     def spin(self):
         for stream in (self.queue_stream, self.notifier_stream, 
                         self.task_stream, self.control_stream):
             stream.flush()
 
 __all__ = [ 'Client', 
             'depend', 
             'require', 
             'remote',
             'parallel',
             'RemoteFunction',
             'ParallelFunction',
             'DirectView',
             'LoadBalancedView',
             'AsyncResult',
             'AsyncMapResult'
             ]

IPython/zmq/parallel/controller.py

@@ -1,108 +1,108 b''
 #!/usr/bin/env python
 """The IPython Controller with 0MQ
 This is a collection of one Hub and several Schedulers.
 """
 #-----------------------------------------------------------------------------
 #  Copyright (C) 2010  The IPython Development Team
 #
 #  Distributed under the terms of the BSD License.  The full license is in
 #  the file COPYING, distributed as part of this software.
 #-----------------------------------------------------------------------------
 
 #-----------------------------------------------------------------------------
 # Imports
 #-----------------------------------------------------------------------------
 from __future__ import print_function
 
 import logging
 from multiprocessing import Process
 
 import zmq
 
 # internal:
 from IPython.utils.importstring import import_item
 from IPython.utils.traitlets import Int, Str, Instance, List, Bool
 
 from entry_point import signal_children
 
 
 from scheduler import launch_scheduler
 from hub import Hub, HubFactory
 
 #-----------------------------------------------------------------------------
 # Configurable
 #-----------------------------------------------------------------------------
 
 
 class ControllerFactory(HubFactory):
     """Configurable for setting up a Hub and Schedulers."""
     
     scheme = Str('pure', config=True)
     usethreads = Bool(False, config=True)
     
     # internal
     children = List()
     mq_class = Str('zmq.devices.ProcessMonitoredQueue')
     
     def _usethreads_changed(self, name, old, new):
         self.mq_class = 'zmq.devices.%sMonitoredQueue'%('Thread' if new else 'Process')
         
     def __init__(self, **kwargs):
         super(ControllerFactory, self).__init__(**kwargs)
         self.subconstructors.append(self.construct_schedulers)
     
     def start(self):
         super(ControllerFactory, self).start()
         for child in self.children:
             child.start()
         if not self.usethreads:
             signal_children([ getattr(c, 'launcher', c) for c in self.children ])
         
     
     def construct_schedulers(self):
         children = self.children
         mq = import_item(self.mq_class)
         
         # IOPub relay (in a Process)
         q = mq(zmq.PUB, zmq.SUB, zmq.PUB, 'N/A','iopub')
         q.bind_in(self.client_addrs['iopub'])
         q.bind_out(self.engine_addrs['iopub'])
         q.setsockopt_out(zmq.SUBSCRIBE, '')
         q.connect_mon(self.monitor_url)
         q.daemon=True
         children.append(q)
 
         # Multiplexer Queue (in a Process)
         q = mq(zmq.XREP, zmq.XREP, zmq.PUB, 'in', 'out')
         q.bind_in(self.client_addrs['mux'])
         q.bind_out(self.engine_addrs['mux'])
         q.connect_mon(self.monitor_url)
         q.daemon=True
         children.append(q)
 
         # Control Queue (in a Process)
         q = mq(zmq.XREP, zmq.XREP, zmq.PUB, 'incontrol', 'outcontrol')
         q.bind_in(self.client_addrs['control'])
         q.bind_out(self.engine_addrs['control'])
         q.connect_mon(self.monitor_url)
         q.daemon=True
         children.append(q)
         # Task Queue (in a Process)
         if self.scheme == 'pure':
             self.log.warn("task::using pure XREQ Task scheduler")
             q = mq(zmq.XREP, zmq.XREQ, zmq.PUB, 'intask', 'outtask')
             q.bind_in(self.client_addrs['task'])
             q.bind_out(self.engine_addrs['task'])
             q.connect_mon(self.monitor_url)
             q.daemon=True
             children.append(q)
         elif self.scheme == 'none':
             self.log.warn("task::using no Task scheduler")
             
         else:
-            self.log.warn("task::using Python %s Task scheduler"%self.scheme)
+            self.log.info("task::using Python %s Task scheduler"%self.scheme)
             sargs = (self.client_addrs['task'], self.engine_addrs['task'], self.monitor_url, self.client_addrs['notification'])
-            q = Process(target=launch_scheduler, args=sargs, kwargs = dict(scheme=self.scheme))
+            q = Process(target=launch_scheduler, args=sargs, kwargs = dict(scheme=self.scheme,logname=self.log.name, loglevel=self.log.level))
             q.daemon=True
             children.append(q)
 

IPython/zmq/parallel/dependency.py

@@ -1,111 +1,111 b''
 """Dependency utilities"""
 
 from IPython.external.decorator import decorator
 from error import UnmetDependency
 
 
 # flags
 ALL = 1 << 0
 ANY = 1 << 1
 HERE = 1 << 2
 ANYWHERE = 1 << 3
 
 
 class depend(object):
     """Dependency decorator, for use with tasks."""
     def __init__(self, f, *args, **kwargs):
         self.f = f
         self.args = args
         self.kwargs = kwargs
     
     def __call__(self, f):
         return dependent(f, self.f, *self.args, **self.kwargs)
 
 class dependent(object):
     """A function that depends on another function.
     This is an object to prevent the closure used
     in traditional decorators, which are not picklable.
     """
 
     def __init__(self, f, df, *dargs, **dkwargs):
         self.f = f
         self.func_name = getattr(f, '__name__', 'f')
         self.df = df
         self.dargs = dargs
         self.dkwargs = dkwargs
 
     def __call__(self, *args, **kwargs):
         if self.df(*self.dargs, **self.dkwargs) is False:
             raise UnmetDependency()
         return self.f(*args, **kwargs)
     
     @property
     def __name__(self):
         return self.func_name
 
 def _require(*names):
     for name in names:
         try:
             __import__(name)
         except ImportError:
             return False
     return True
 
 def require(*names):
     return depend(_require, *names)
 
 class Dependency(set):
-    """An object for representing a set of dependencies.
+    """An object for representing a set of msg_id dependencies.
     
     Subclassed from set()."""
     
     mode='all'
     success_only=True
     
     def __init__(self, dependencies=[], mode='all', success_only=True):
         if isinstance(dependencies, dict):
             # load from dict
             mode = dependencies.get('mode', mode)
             success_only = dependencies.get('success_only', success_only)
             dependencies = dependencies.get('dependencies', [])
         set.__init__(self, dependencies)
         self.mode = mode.lower()
         self.success_only=success_only
         if self.mode not in ('any', 'all'):
             raise NotImplementedError("Only any|all supported, not %r"%mode)
     
     def check(self, completed, failed=None):
         if failed is not None and not self.success_only:
             completed = completed.union(failed)
         if len(self) == 0:
             return True
         if self.mode == 'all':
             return self.issubset(completed)
         elif self.mode == 'any':
             return not self.isdisjoint(completed)
         else:
             raise NotImplementedError("Only any|all supported, not %r"%mode)
     
     def unreachable(self, failed):
         if len(self) == 0 or len(failed) == 0 or not self.success_only:
             return False
         print self, self.success_only, self.mode, failed
         if self.mode == 'all':
             return not self.isdisjoint(failed)
         elif self.mode == 'any':
             return self.issubset(failed)
         else:
             raise NotImplementedError("Only any|all supported, not %r"%mode)
         
     
     def as_dict(self):
         """Represent this dependency as a dict. For json compatibility."""
         return dict(
             dependencies=list(self),
             mode=self.mode,
             success_only=self.success_only,
         )
     
 
 __all__ = ['depend', 'require', 'Dependency']
 

IPython/zmq/parallel/error.py

@@ -1,289 +1,292 b''
 # encoding: utf-8
 
 """Classes and functions for kernel related errors and exceptions."""
 from __future__ import print_function
 
 __docformat__ = "restructuredtext en"
 
 # Tell nose to skip this module
 __test__ = {}
 
 #-------------------------------------------------------------------------------
 #  Copyright (C) 2008  The IPython Development Team
 #
 #  Distributed under the terms of the BSD License.  The full license is in
 #  the file COPYING, distributed as part of this software.
 #-------------------------------------------------------------------------------
 
 #-------------------------------------------------------------------------------
 # Error classes
 #-------------------------------------------------------------------------------
 class IPythonError(Exception):
     """Base exception that all of our exceptions inherit from.
 
     This can be raised by code that doesn't have any more specific
     information."""
 
     pass
 
 # Exceptions associated with the controller objects
 class ControllerError(IPythonError): pass
 
 class ControllerCreationError(ControllerError): pass
 
 
 # Exceptions associated with the Engines
 class EngineError(IPythonError): pass
 
 class EngineCreationError(EngineError): pass
 
 class KernelError(IPythonError):
     pass
 
 class NotDefined(KernelError):
     def __init__(self, name):
         self.name = name
         self.args = (name,)
 
     def __repr__(self):
         return '<NotDefined: %s>' % self.name
     
     __str__ = __repr__
 
 
 class QueueCleared(KernelError):
     pass
 
 
 class IdInUse(KernelError):
     pass
 
 
 class ProtocolError(KernelError):
     pass
 
 
 class ConnectionError(KernelError):
     pass
 
 
 class InvalidEngineID(KernelError):
     pass
 
 
 class NoEnginesRegistered(KernelError):
     pass
 
 
 class InvalidClientID(KernelError):
     pass
 
 
 class InvalidDeferredID(KernelError):
     pass
 
 
 class SerializationError(KernelError):
     pass
 
 
 class MessageSizeError(KernelError):
     pass
 
 
 class PBMessageSizeError(MessageSizeError):
     pass
 
 
 class ResultNotCompleted(KernelError):
     pass
 
 
 class ResultAlreadyRetrieved(KernelError):
     pass
 
 class ClientError(KernelError):
     pass
 
 
 class TaskAborted(KernelError):
     pass
 
 
 class TaskTimeout(KernelError):
     pass
 
 
 class NotAPendingResult(KernelError):
     pass
 
 
 class UnpickleableException(KernelError):
     pass
 
 
 class AbortedPendingDeferredError(KernelError):
     pass
 
 
 class InvalidProperty(KernelError):
     pass
 
 
 class MissingBlockArgument(KernelError):
     pass
 
 
 class StopLocalExecution(KernelError):
     pass
 
 
 class SecurityError(KernelError):
     pass
 
 
 class FileTimeoutError(KernelError):
     pass
 
 class TimeoutError(KernelError):
     pass
 
 class UnmetDependency(KernelError):
     pass
 
 class ImpossibleDependency(UnmetDependency):
     pass
 
+class DependencyTimeout(UnmetDependency):
+    pass
+
 class RemoteError(KernelError):
     """Error raised elsewhere"""
     ename=None
     evalue=None
     traceback=None
     engine_info=None
     
     def __init__(self, ename, evalue, traceback, engine_info=None):
         self.ename=ename
         self.evalue=evalue
         self.traceback=traceback
         self.engine_info=engine_info or {}
         self.args=(ename, evalue)
     
     def __repr__(self):
         engineid = self.engine_info.get('engineid', ' ')
         return "<Remote[%s]:%s(%s)>"%(engineid, self.ename, self.evalue)
         
     def __str__(self):
         sig = "%s(%s)"%(self.ename, self.evalue)
         if self.traceback:
             return sig + '\n' + self.traceback
         else:
             return sig
     
 
 class TaskRejectError(KernelError):
     """Exception to raise when a task should be rejected by an engine.
     
     This exception can be used to allow a task running on an engine to test
     if the engine (or the user's namespace on the engine) has the needed
     task dependencies.  If not, the task should raise this exception.  For
     the task to be retried on another engine, the task should be created
     with the `retries` argument > 1.
     
     The advantage of this approach over our older properties system is that
     tasks have full access to the user's namespace on the engines and the
     properties don't have to be managed or tested by the controller.
     """
 
 
 class CompositeError(KernelError):
     """Error for representing possibly multiple errors on engines"""
     def __init__(self, message, elist):
         Exception.__init__(self, *(message, elist))
         # Don't use pack_exception because it will conflict with the .message
         # attribute that is being deprecated in 2.6 and beyond.
         self.msg = message
         self.elist = elist
         self.args = [ e[0] for e in elist ]
 
     def _get_engine_str(self, ei):
         if not ei:
             return '[Engine Exception]'
         else:
             return '[%i:%s]: ' % (ei['engineid'], ei['method'])
 
     def _get_traceback(self, ev):
         try:
             tb = ev._ipython_traceback_text
         except AttributeError:
             return 'No traceback available'
         else:
             return tb
 
     def __str__(self):
         s = str(self.msg)
         for en, ev, etb, ei in self.elist:
             engine_str = self._get_engine_str(ei)
             s = s + '\n' + engine_str + en + ': ' + str(ev)
         return s
     
     def __repr__(self):
         return "CompositeError(%i)"%len(self.elist)
 
     def print_tracebacks(self, excid=None):
         if excid is None:
             for (en,ev,etb,ei) in self.elist:
                 print (self._get_engine_str(ei))
                 print (etb or 'No traceback available')
                 print ()
         else:
             try:
                 en,ev,etb,ei = self.elist[excid]
             except:
                 raise IndexError("an exception with index %i does not exist"%excid)
             else:
                 print (self._get_engine_str(ei))
                 print (etb or 'No traceback available')
     
     def raise_exception(self, excid=0):
         try:
             en,ev,etb,ei = self.elist[excid]
         except:
             raise IndexError("an exception with index %i does not exist"%excid)
         else:
             try:
                 raise RemoteError(en, ev, etb, ei)
             except:
                 et,ev,tb = sys.exc_info()
 
 
 def collect_exceptions(rdict_or_list, method='unspecified'):
     """check a result dict for errors, and raise CompositeError if any exist.
     Passthrough otherwise."""
     elist = []
     if isinstance(rdict_or_list, dict):
         rlist = rdict_or_list.values()
     else:
         rlist = rdict_or_list
     for r in rlist:
         if isinstance(r, RemoteError):
             en, ev, etb, ei = r.ename, r.evalue, r.traceback, r.engine_info
             # Sometimes we could have CompositeError in our list.  Just take
             # the errors out of them and put them in our new list.  This 
             # has the effect of flattening lists of CompositeErrors into one
             # CompositeError
             if en=='CompositeError':
                 for e in ev.elist:
                     elist.append(e)
             else:
                 elist.append((en, ev, etb, ei))
     if len(elist)==0:
         return rdict_or_list
     else:
         msg = "one or more exceptions from call to method: %s" % (method)
         # This silliness is needed so the debugger has access to the exception
         # instance (e in this case)
         try:
             raise CompositeError(msg, elist)
         except CompositeError, e:
             raise e
 

IPython/zmq/parallel/scheduler.py

@@ -1,511 +1,526 b''
 """The Python scheduler for rich scheduling.
 
 The Pure ZMQ scheduler does not allow routing schemes other than LRU,
 nor does it check msg_id DAG dependencies. For those, a slightly slower
 Python Scheduler exists.
 """
 
 #----------------------------------------------------------------------
 # Imports
 #----------------------------------------------------------------------
 
 from __future__ import print_function
 import sys
 import logging
-from random import randint,random
+from random import randint, random
 from types import FunctionType
-
+from datetime import datetime, timedelta
 try:
     import numpy
 except ImportError:
     numpy = None
 
 import zmq
 from zmq.eventloop import ioloop, zmqstream
 
 # local imports
 from IPython.external.decorator import decorator
 # from IPython.config.configurable import Configurable
 from IPython.utils.traitlets import Instance, Dict, List, Set
 
 import error
-from client import Client
+# from client import Client
 from dependency import Dependency
 import streamsession as ss
 from entry_point import connect_logger, local_logger
-from factory import LoggingFactory
+from factory import SessionFactory
 
 
 @decorator
 def logged(f,self,*args,**kwargs):
     # print ("#--------------------")
     self.log.debug("scheduler::%s(*%s,**%s)"%(f.func_name, args, kwargs))
     # print ("#--")
     return f(self,*args, **kwargs)
 
 #----------------------------------------------------------------------
 # Chooser functions
 #----------------------------------------------------------------------
 
 def plainrandom(loads):
     """Plain random pick."""
     n = len(loads)
     return randint(0,n-1)
 
 def lru(loads):
     """Always pick the front of the line.
     
     The content of `loads` is ignored.
     
     Assumes LRU ordering of loads, with oldest first.
     """
     return 0
 
 def twobin(loads):
     """Pick two at random, use the LRU of the two.
     
     The content of loads is ignored.
     
     Assumes LRU ordering of loads, with oldest first.
     """
     n = len(loads)
     a = randint(0,n-1)
     b = randint(0,n-1)
     return min(a,b)
 
 def weighted(loads):
     """Pick two at random using inverse load as weight.
     
     Return the less loaded of the two.
     """
     # weight 0 a million times more than 1:
     weights = 1./(1e-6+numpy.array(loads))
     sums = weights.cumsum()
     t = sums[-1]
     x = random()*t
     y = random()*t
     idx = 0
     idy = 0
     while sums[idx] < x:
         idx += 1
     while sums[idy] < y:
         idy += 1
     if weights[idy] > weights[idx]:
         return idy
     else:
         return idx
 
 def leastload(loads):
     """Always choose the lowest load.
     
     If the lowest load occurs more than once, the first
     occurance will be used.  If loads has LRU ordering, this means
     the LRU of those with the lowest load is chosen.
     """
     return loads.index(min(loads))
 
 #---------------------------------------------------------------------
 # Classes
 #---------------------------------------------------------------------
 # store empty default dependency:
 MET = Dependency([])
 
-class TaskScheduler(LoggingFactory):
+class TaskScheduler(SessionFactory):
     """Python TaskScheduler object.
     
     This is the simplest object that supports msg_id based
     DAG dependencies. *Only* task msg_ids are checked, not
     msg_ids of jobs submitted via the MUX queue.
     
     """
     
     # input arguments:
     scheme = Instance(FunctionType, default=leastload) # function for determining the destination
     client_stream = Instance(zmqstream.ZMQStream) # client-facing stream
     engine_stream = Instance(zmqstream.ZMQStream) # engine-facing stream
     notifier_stream = Instance(zmqstream.ZMQStream) # hub-facing sub stream
     mon_stream = Instance(zmqstream.ZMQStream) # hub-facing pub stream
-    io_loop = Instance(ioloop.IOLoop)
     
     # internals:
     dependencies = Dict() # dict by msg_id of [ msg_ids that depend on key ]
     depending = Dict() # dict by msg_id of (msg_id, raw_msg, after, follow)
     pending = Dict() # dict by engine_uuid of submitted tasks
     completed = Dict() # dict by engine_uuid of completed tasks
     failed = Dict() # dict by engine_uuid of failed tasks
     destinations = Dict() # dict by msg_id of engine_uuids where jobs ran (reverse of completed+failed)
     clients = Dict() # dict by msg_id for who submitted the task
     targets = List() # list of target IDENTs
     loads = List() # list of engine loads
     all_completed = Set() # set of all completed tasks
     all_failed = Set() # set of all failed tasks
     all_done = Set() # set of all finished tasks=union(completed,failed)
     blacklist = Dict() # dict by msg_id of locations where a job has encountered UnmetDependency
-    session = Instance(ss.StreamSession)
+    auditor = Instance('zmq.eventloop.ioloop.PeriodicCallback')
     
     
-    def __init__(self, **kwargs):
-        super(TaskScheduler, self).__init__(**kwargs)
-        
-        self.session = ss.StreamSession(username="TaskScheduler")
-        
+    def start(self):
         self.engine_stream.on_recv(self.dispatch_result, copy=False)
         self._notification_handlers = dict(
             registration_notification = self._register_engine,
             unregistration_notification = self._unregister_engine
         )
         self.notifier_stream.on_recv(self.dispatch_notification)
+        self.auditor = ioloop.PeriodicCallback(self.audit_timeouts, 1e3, self.loop) # 1 Hz
+        self.auditor.start()
         self.log.info("Scheduler started...%r"%self)
     
     def resume_receiving(self):
         """Resume accepting jobs."""
         self.client_stream.on_recv(self.dispatch_submission, copy=False)
     
     def stop_receiving(self):
         """Stop accepting jobs while there are no engines.
         Leave them in the ZMQ queue."""
         self.client_stream.on_recv(None)
     
     #-----------------------------------------------------------------------
     # [Un]Registration Handling
     #-----------------------------------------------------------------------
     
     def dispatch_notification(self, msg):
         """dispatch register/unregister events."""
         idents,msg = self.session.feed_identities(msg)
         msg = self.session.unpack_message(msg)
         msg_type = msg['msg_type']
         handler = self._notification_handlers.get(msg_type, None)
         if handler is None:
             raise Exception("Unhandled message type: %s"%msg_type)
         else:
             try:
                 handler(str(msg['content']['queue']))
             except KeyError:
                 self.log.error("task::Invalid notification msg: %s"%msg)
     
     @logged
     def _register_engine(self, uid):
         """New engine with ident `uid` became available."""
         # head of the line:
         self.targets.insert(0,uid)
         self.loads.insert(0,0)
         # initialize sets
         self.completed[uid] = set()
         self.failed[uid] = set()
         self.pending[uid] = {}
         if len(self.targets) == 1:
             self.resume_receiving()
 
     def _unregister_engine(self, uid):
         """Existing engine with ident `uid` became unavailable."""
         if len(self.targets) == 1:
             # this was our only engine
             self.stop_receiving()
         
         # handle any potentially finished tasks:
         self.engine_stream.flush()
         
         self.completed.pop(uid)
         self.failed.pop(uid)
         # don't pop destinations, because it might be used later
         # map(self.destinations.pop, self.completed.pop(uid))
         # map(self.destinations.pop, self.failed.pop(uid))
         
         lost = self.pending.pop(uid)
         
         idx = self.targets.index(uid)
         self.targets.pop(idx)
         self.loads.pop(idx)
         
         self.handle_stranded_tasks(lost)
     
     def handle_stranded_tasks(self, lost):
         """Deal with jobs resident in an engine that died."""
         # TODO: resubmit the tasks?
         for msg_id in lost:
             pass
     
     
     #-----------------------------------------------------------------------
     # Job Submission
     #-----------------------------------------------------------------------
     @logged
     def dispatch_submission(self, raw_msg):
         """Dispatch job submission to appropriate handlers."""
         # ensure targets up to date:
         self.notifier_stream.flush()
         try:
             idents, msg = self.session.feed_identities(raw_msg, copy=False)
         except Exception as e:
             self.log.error("task::Invaid msg: %s"%msg)
             return
         
         # send to monitor
         self.mon_stream.send_multipart(['intask']+raw_msg, copy=False)
         
         msg = self.session.unpack_message(msg, content=False, copy=False)
         header = msg['header']
         msg_id = header['msg_id']
         
         # time dependencies
         after = Dependency(header.get('after', []))
         if after.mode == 'all':
             after.difference_update(self.all_completed)
             if not after.success_only:
                 after.difference_update(self.all_failed)
         if after.check(self.all_completed, self.all_failed):
             # recast as empty set, if `after` already met,
             # to prevent unnecessary set comparisons
             after = MET
         
         # location dependencies
         follow = Dependency(header.get('follow', []))
-        
         # check if unreachable:
         if after.unreachable(self.all_failed) or follow.unreachable(self.all_failed):
-            self.depending[msg_id] = [raw_msg,MET,MET]
+            self.depending[msg_id] = [raw_msg,MET,MET,None]
             return self.fail_unreachable(msg_id)
         
+        # turn timeouts into datetime objects:
+        timeout = header.get('timeout', None)
+        if timeout:
+            timeout = datetime.now() + timedelta(0,timeout,0)
+        
         if after.check(self.all_completed, self.all_failed):
             # time deps already met, try to run
             if not self.maybe_run(msg_id, raw_msg, follow):
                 # can't run yet
-                self.save_unmet(msg_id, raw_msg, after, follow)
+                self.save_unmet(msg_id, raw_msg, after, follow, timeout)
         else:
-            self.save_unmet(msg_id, raw_msg, after, follow)
+            self.save_unmet(msg_id, raw_msg, after, follow, timeout)
     
     @logged
-    def fail_unreachable(self, msg_id):
+    def audit_timeouts(self):
+        """Audit all waiting tasks for expired timeouts."""
+        now = datetime.now()
+        for msg_id in self.depending.keys():
+            # must recheck, in case one failure cascaded to another:
+            if msg_id in self.depending:
+                raw,after,follow,timeout = self.depending[msg_id]
+                if timeout and timeout < now:
+                    self.fail_unreachable(msg_id, timeout=True)
+                
+    @logged
+    def fail_unreachable(self, msg_id, timeout=False):
         """a message has become unreachable"""
         if msg_id not in self.depending:
             self.log.error("msg %r already failed!"%msg_id)
             return
-        raw_msg, after, follow = self.depending.pop(msg_id)
+        raw_msg, after, follow, timeout = self.depending.pop(msg_id)
         for mid in follow.union(after):
             if mid in self.dependencies:
                 self.dependencies[mid].remove(msg_id)
         
+        # FIXME: unpacking a message I've already unpacked, but didn't save:
         idents,msg = self.session.feed_identities(raw_msg, copy=False)
         msg = self.session.unpack_message(msg, copy=False, content=False)
         header = msg['header']
         
+        impossible = error.DependencyTimeout if timeout else error.ImpossibleDependency
+        
         try:
-            raise error.ImpossibleDependency()
+            raise impossible()
         except:
             content = ss.wrap_exception()
         
         self.all_done.add(msg_id)
         self.all_failed.add(msg_id)
         
         msg = self.session.send(self.client_stream, 'apply_reply', content, 
                                                 parent=header, ident=idents)
         self.session.send(self.mon_stream, msg, ident=['outtask']+idents)
         
         self.update_dependencies(msg_id, success=False)
     
     @logged
     def maybe_run(self, msg_id, raw_msg, follow=None):
         """check location dependencies, and run if they are met."""
             
         if follow:
             def can_run(idx):
                 target = self.targets[idx]
                 return target not in self.blacklist.get(msg_id, []) and\
                         follow.check(self.completed[target], self.failed[target])
             
             indices = filter(can_run, range(len(self.targets)))
             if not indices:
                 # TODO evaluate unmeetable follow dependencies
                 if follow.mode == 'all':
                     dests = set()
                     relevant = self.all_completed if follow.success_only else self.all_done
                     for m in follow.intersection(relevant):
                         dests.add(self.destinations[m])
                     if len(dests) > 1:
                         self.fail_unreachable(msg_id)
                 
                 
                 return False
         else:
             indices = None
             
         self.submit_task(msg_id, raw_msg, indices)
         return True
             
     @logged
-    def save_unmet(self, msg_id, raw_msg, after, follow):
+    def save_unmet(self, msg_id, raw_msg, after, follow, timeout):
         """Save a message for later submission when its dependencies are met."""
-        self.depending[msg_id] = [raw_msg,after,follow]
+        self.depending[msg_id] = [raw_msg,after,follow,timeout]
         # track the ids in follow or after, but not those already finished
         for dep_id in after.union(follow).difference(self.all_done):
             if dep_id not in self.dependencies:
                 self.dependencies[dep_id] = set()
             self.dependencies[dep_id].add(msg_id)
     
     @logged
     def submit_task(self, msg_id, msg, follow=None, indices=None):
         """Submit a task to any of a subset of our targets."""
         if indices:
             loads = [self.loads[i] for i in indices]
         else:
             loads = self.loads
         idx = self.scheme(loads)
         if indices:
             idx = indices[idx]
         target = self.targets[idx]
         # print (target, map(str, msg[:3]))
         self.engine_stream.send(target, flags=zmq.SNDMORE, copy=False)
         self.engine_stream.send_multipart(msg, copy=False)
         self.add_job(idx)
         self.pending[target][msg_id] = (msg, follow)
         content = dict(msg_id=msg_id, engine_id=target)
         self.session.send(self.mon_stream, 'task_destination', content=content, 
                         ident=['tracktask',self.session.session])
     
     #-----------------------------------------------------------------------
     # Result Handling
     #-----------------------------------------------------------------------
     @logged
     def dispatch_result(self, raw_msg):
         try:
             idents,msg = self.session.feed_identities(raw_msg, copy=False)
         except Exception as e:
             self.log.error("task::Invaid result: %s"%msg)
             return
         msg = self.session.unpack_message(msg, content=False, copy=False)
         header = msg['header']
         if header.get('dependencies_met', True):
             success = (header['status'] == 'ok')
             self.handle_result(idents, msg['parent_header'], raw_msg, success)
             # send to Hub monitor
             self.mon_stream.send_multipart(['outtask']+raw_msg, copy=False)
         else:
             self.handle_unmet_dependency(idents, msg['parent_header'])
         
     @logged
     def handle_result(self, idents, parent, raw_msg, success=True):
         # first, relay result to client
         engine = idents[0]
         client = idents[1]
         # swap_ids for XREP-XREP mirror
         raw_msg[:2] = [client,engine]
         # print (map(str, raw_msg[:4]))
         self.client_stream.send_multipart(raw_msg, copy=False)
         # now, update our data structures
         msg_id = parent['msg_id']
         self.pending[engine].pop(msg_id)
         if success:
             self.completed[engine].add(msg_id)
             self.all_completed.add(msg_id)
         else:
             self.failed[engine].add(msg_id)
             self.all_failed.add(msg_id)
         self.all_done.add(msg_id)
         self.destinations[msg_id] = engine
         
         self.update_dependencies(msg_id, success)
         
     @logged
     def handle_unmet_dependency(self, idents, parent):
         engine = idents[0]
         msg_id = parent['msg_id']
         if msg_id not in self.blacklist:
             self.blacklist[msg_id] = set()
         self.blacklist[msg_id].add(engine)
-        raw_msg,follow = self.pending[engine].pop(msg_id)
+        raw_msg,follow,timeout = self.pending[engine].pop(msg_id)
         if not self.maybe_run(msg_id, raw_msg, follow):
             # resubmit failed, put it back in our dependency tree
-            self.save_unmet(msg_id, raw_msg, MET, follow)
+            self.save_unmet(msg_id, raw_msg, MET, follow, timeout)
         pass
     
     @logged
     def update_dependencies(self, dep_id, success=True):
         """dep_id just finished. Update our dependency
         table and submit any jobs that just became runable."""
         # print ("\n\n***********")
         # pprint (dep_id)
         # pprint (self.dependencies)
         # pprint (self.depending)
         # pprint (self.all_completed)
         # pprint (self.all_failed)
         # print ("\n\n***********\n\n")
         if dep_id not in self.dependencies:
             return
         jobs = self.dependencies.pop(dep_id)
         
         for msg_id in jobs:
-            raw_msg, after, follow = self.depending[msg_id]
+            raw_msg, after, follow, timeout = self.depending[msg_id]
             # if dep_id in after:
             #     if after.mode == 'all' and (success or not after.success_only):
             #         after.remove(dep_id)
             
             if after.unreachable(self.all_failed) or follow.unreachable(self.all_failed):
                 self.fail_unreachable(msg_id)
             
             elif after.check(self.all_completed, self.all_failed): # time deps met, maybe run
                 self.depending[msg_id][1] = MET
                 if self.maybe_run(msg_id, raw_msg, follow):
                     
                     self.depending.pop(msg_id)
                     for mid in follow.union(after):
                         if mid in self.dependencies:
                             self.dependencies[mid].remove(msg_id)
     
     #----------------------------------------------------------------------
     # methods to be overridden by subclasses
     #----------------------------------------------------------------------
     
     def add_job(self, idx):
         """Called after self.targets[idx] just got the job with header.
         Override with subclasses.  The default ordering is simple LRU.
         The default loads are the number of outstanding jobs."""
         self.loads[idx] += 1
         for lis in (self.targets, self.loads):
             lis.append(lis.pop(idx))
             
     
     def finish_job(self, idx):
         """Called after self.targets[idx] just finished a job.
         Override with subclasses."""
         self.loads[idx] -= 1
     
 
 
 def launch_scheduler(in_addr, out_addr, mon_addr, not_addr, logname='ZMQ', log_addr=None, loglevel=logging.DEBUG, scheme='weighted'):
     from zmq.eventloop import ioloop
     from zmq.eventloop.zmqstream import ZMQStream
     
     ctx = zmq.Context()
     loop = ioloop.IOLoop()
     
     ins = ZMQStream(ctx.socket(zmq.XREP),loop)
     ins.bind(in_addr)
     outs = ZMQStream(ctx.socket(zmq.XREP),loop)
     outs.bind(out_addr)
     mons = ZMQStream(ctx.socket(zmq.PUB),loop)
     mons.connect(mon_addr)
     nots = ZMQStream(ctx.socket(zmq.SUB),loop)
     nots.setsockopt(zmq.SUBSCRIBE, '')
     nots.connect(not_addr)
     
     scheme = globals().get(scheme, None)
     # setup logging
     if log_addr:
         connect_logger(logname, ctx, log_addr, root="scheduler", loglevel=loglevel)
     else:
         local_logger(logname, loglevel)
     
     scheduler = TaskScheduler(client_stream=ins, engine_stream=outs,
-                            mon_stream=mons,notifier_stream=nots,
-                            scheme=scheme,io_loop=loop, logname=logname)
-    
+                            mon_stream=mons, notifier_stream=nots,
+                            scheme=scheme, loop=loop, logname=logname)
+    scheduler.start()
     try:
         loop.start()
     except KeyboardInterrupt:
         print ("interrupted, exiting...", file=sys.__stderr__)
 
 
 if __name__ == '__main__':
     iface = 'tcp://127.0.0.1:%i'
     launch_scheduler(iface%12345,iface%1236,iface%12347,iface%12348)