Decisions from the 2009 Ganga dev. days in Oslo

It's agreed that the Ganga Core framework should provide a basic class for the Thread objects which may be created in Ganga plug-ins for multi-threads operations. The basic class should implements proper methods by which the thread objects derived from the basic class can be managed and controlled centrally in a thread pool.

Implementation

Implementation is done in the Ganga.Core.GangaThread module containing the following classes:

  • GangaThreadPool: a singelton maintaining a list of thread objects derived from the GangaThread class
  • GangaThread: the basic class extended from threading.Thread. All thread object in Ganga should be inherited from it.

Concerning of handling parametric-sweep-like activities in multiple parallel threads, a helper module called MTRunner is also provided based on a pooling thread approach. The module consists of 3 main objects:

  • MTRunner: the multi-thread activity runner
  • Algorithm: the object for defining the runtime action
  • Data: the data object defining the parameters or datasets the runtime action should execute on

Usage

Using the GangaThread basic class

The following example shows how to define your own thread object extending the GangaThread object:

from Ganga.Core.GangaThread import GangaThread
from Ganga.Utility.logging import getLogger

class MyThread(GangaThread):

    def __init__(self):
        GangaThread.__init__(self, name='my_thread')
        self.logger = getLogger()

    def run(self):
    
        ## run a loop until the thread is notified to stop
        ## the "should stop" notification comes from the Ganga shutdown service     
        while not self.should_stop():
            logger.info('still working on it ...')

        ## unregistered the thread from the GangaThreadPool as it's going to finish
        self.unregister()

and in the main thread, just create and run MyThread as the following:

... ...
t = MyThread()
t.start()
... ...

Using the MTRunner

The following example demonstrates how to use MTRunner to say "Hello" to multiple people in parallel.

Define the runtime action to print out the customized greeting messages:

from Ganga.Core.GangaThread.MTRunner import Algorithm

class GreetingAlgorithm(Algorithm):
    def process(self, item):
        print 'Hello, %s!' % item
        return True

Define the activity runner:

from Ganga.Core.GangaThread.MTRunner import MTRunner

class Greeter(MTRunner):

    def __init__(self, numThread, keepAlive):
        MTRunner.__init__(self, name='my_greeter', data=Data(collection=[]), algorithm=GreetingAlgorithm())

        ## specify number of agent threads to be created (i.e. number of parallel threads)
        self.numThread = numThread

        ## specify if the agent threads should be kept alive in the background
        self.keepAlive = keepAlive

    ## a method allow adding new people on demand
    def addNewPerson(self, name):
        self.addDataItem(name)

Run the Greeter

## here we ask for 10 greeting agents and let the agents alive even there is no data to process
g = Greeter(numThread=10, keepAlive=True)

## add some initial data to be processed (people to be greeted)
g.addNewPerson(name="David")
g.addNewPerson(name="Hellen")
g.run()

## here you can run something else in your main thread and the greeter to take care of the greeting stuff
... ...

## now add more new persons
g.addNewPerson(name="Jean")

-- HurngChunLee - 12 Mar 2009

Edit | Attach | Watch | Print version | History: r1 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r1 - 2009-03-12 - HurngChunLee
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    ArdaGrid All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2022 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
or Ideas, requests, problems regarding TWiki? use Discourse or Send feedback