About

License

Simplified BSD:

Copyright (c) since 2006, Oliver Schoenborn
All rights reserved.

Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are met: 

1. Redistributions of source code must retain the above copyright notice, this
   list of conditions and the following disclaimer. 
2. Redistributions in binary form must reproduce the above copyright notice,
   this list of conditions and the following disclaimer in the documentation
   and/or other materials provided with the distribution. 

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

Authors

The main developer of the package is Oliver Schoenborn, but Robb Shecter started it all back in early 2000’s. The code is now hosted on github at http://github.com/schollii/pypubsub (previous to that, it was hosted on SourceForge.net, and prior to that, it was hosted in wxPython). PyPubSub is on the Python Package Index at http://pypi.python.org/pypi/PyPubSub.

As listed in the Contributing section, contributions of any form are welcome. Even questions help progress the project.

author:

Oliver Schoenborn <schoenborno at (@) users.sf.net>

Acknowledgements

Thanks to SourceForge.net for hosting the project until September 2016. Thanks to Github.com for hosting the project from October 2016. Thanks to Robb Shecter for having given me the chance to take over this project from him many years ago (ca 2004!). Thanks to all those users of PyPubSub who ask questions, make suggestions, point out bugs, etc.

PyPubSub Users

The page at https://libraries.io/pypi/PyPubSub has useful statics on which other github projects use/watch/fork pypubsub.

In addition, several users have donated a bit of their time to describe how they use/d PyPubSub in their Python projects.

Darin Gordon, for Yosai, since 2015:

Yosai (https://github.com/YosaiProject) is a security framework for python applications, offering authentication, authorization, and session management from a common API. Yosai uses PyPubSub to facilitate event-driven responses to security-related changes. For instance, when a session expires, valuable information is logged and cached authorization info is cleared.

Jerome Laheurte, for Task Coach, since Feb 2012:

Task Coach (https://sourceforge.net/projects/taskcoach/) is a simple open source todo manager to keep track of personal tasks and todo lists. It is designed for composite tasks, and also offers effort tracking, categories, notes and more. Task Coach uses PyPubSub as its Publisher/Listener implementation to cleanly separate model and view layers.

Steven Sproat, for Whyteboard, since Feb 2010:

I’ve been using PyPubSub for around 2 months in my cross-platform drawing application, Whyteboard (http://launchpad.net/whyteboard). My Shape models (rectangle, polygons etc) use PyPubSub to notify the GUI of any changes to themselves or to request actions be performed on the canvas (e.g capture user’s mouse), and the GUI responds by updating various dialogs with this information. This means that my shapes no longer need to maintain references to the canvas in order to perform operations on it, and can instead send a message saying “do something” without caring how it’s done.

Josh English, for WMS, since 2008:

I use it in my Writing Management System (http://joshua.r.english .googlepages.com/wms). I’m using it to control interfaces, such as telling the frame to change the status bar, or a notebook to change a panel. PyPubSub enables me to focus on what data to pass around my application, rather than how to pass it around. This makes it easy to put in the finer details of my application.

Geoff Gilmour-Taylor, since April 2008:

I use wx.lib.pubsub for a suite of in-house batch conversion tools for DAISY talking books, called Garden Tools (in-house software for the CNIB Library, http://www.cnib.ca/en/Services/library/). For MVC, communication in a wxPython app.

Loose coupling of business logic and GUI. It allows me to trigger multiple actions on a single message without having to locate and modify all the places where the message is sent. I was able to add a logging module that reads the same status messages that are sent to the GUI without having to modify any of my other code.

Phil Mayes, for Listomax, since 2007:

Listomax (http://www.listomax.com/) uses version 1 of PyPubSub for MVC: multiple View (UI) components may need to change when the Model changes; simpler than direct calls, lower coupling.

Mike Driscoll, for PyTimesheet and Zimbra Alerts, since 2007:

I use wx.lib.pubsub in two internal projects at my employer’s business, “PyTimesheet” and the “Zimbra Alerts”. I use it to send information between various frames, such as an options menu back to the main application that launched it. The main application I use it for though is a Timesheet program where I use it to tell my program which frame to display when. Basically when one closes, I need another one to open and I found that PyPubSub made this quite trivial. The other program is used in conjunction with our Zimbra web mail and will pop-up an alert when we receive an email and it also has an Outlook-like Reminder dialog for appointments… And thanks for providing such a nice tool for my arsenal!

Anthony Floyd, RAVEN, since 200?:

Our project is called “RAVEN”, it’s an analytical and finite-element analysis program for simulating and analyzing the processing of composite materials in the aerospace industry. We use PyPubSub as the communications backbone. We essentially have a MVC framework, and use PyPubSub to have the UI respond to things happening in the data. However, we also use it to have data objects respond to changes in other data objects.

We’re quite enamoured with PyPubSub! It’s proven to be an effective way to keep the UI out of the backend, and an effective way to keep the backend modularized.

Sebastian Zurek, OpenSynergy, since 2007:

I’m using wx.lib.pubsub module as part of the OpenSynergy framework (http://www.opensynergy.pl, temporarily offline) that I am developing, and I found it VERY usefull. PyPubSub is used as the communication layer betteen the extensions components and the framework, between the Model and Visual, and between the Visual elements.

Werner F. Bruhin, for the The Wine Cellar Book, since 2006

Have been using PyPubSub for years and since I started work on version 4 of my application [http://thewinecellarbook.com)] over a year ago I switched to the PyPubSub v3 API and defined a topic tree. Having a topic tree is just great as you make sure that you don’t mistype the topic names and on top you have nice documentation on what topics you already defined and what parameter(s) need to be passed for each topic.

Currently I have topics to keep track of database state, data needs saving, database item/row has changed, was deleted etc which trigger updates to relevant lists to update themselves if needed and to show messages on a wx.InfoBar.

Mike Rooney, for wxBanker, since 2006:

I use PyPubSub as the crucial event handling mechanism for wxBanker (https://launchpad.net/wxbanker). It works well for implementing design patterns such as MVC where you want to eliminate coupling, since it doesn’t require that you know specific method names or implementation details of other classes, modules, or libraries. PyPubSub is also great when you want to make an announcement without requiring that anything (or how many things) is listening to or acting upon that announcement. In short, PyPubSub makes intra-process communication a dream come true.

QVI (http://www.qvii.com/) for several applications, since 2006:

Here at QVI we use PyPubSub for most of our wxPython applications (notably SmartTree), to achieve very lightweight, simple, and readable communication between classes and modules. One of the nice aspects of PyPubSub is how easy it is to incorporate into existing code, and how well-suited it is for pluggable/modular designs which want to make announcements about events, but don’t require that or care if any other module is listening. It makes handling “events” easy, whatever we define them to be, and removes the need for the handlers to have any specific knowledge of how the announcements are made or where they came from.

After discovering we could use PyPubSub independently of wxPython, we also use it in an application or two that doesn’t use wxPython at all, but where we still desire a lightweight event handling mechanism (when don’t you?).

Oliver Schoenborn (Author of PyPubSub), for several applications, from 2004 to 2010:

I have used PyPubSub on several projects. Applications which, for example,

  • show tree structures with selectable nodes, and selected node’s associated information panel

  • show objects on information panels with info regarding progress of other components (running on other machines) updated in real time

  • show dialog boxes with with entry fields for settings

  • have several panels in a wizard style to configure a task for execution

With PyPubSub, one event occurs due to a mouse click on an icon, and and all parts of the code that need updating get called with the new data. This means automatic update of menu items (adding, removing etc), state in various panels, etc. Gone are the long sequences of calls through the code.

Last time I had to build or maintain a Python event-based application was 2009, but I’m dedicated to maintaining PyPubSub for other developers. When I make or incorporate improvements based on user feedback, I rely on the high % coverage of the unit regression tests, the useful examples, and the exception messages which give a lot of useful information when the message data fields don’t adhere to the topic definition (inferred or specified).

I look forward to my next event-based Python application, which might be in the fall of 2013.

History

PyPubSub was originally created by Robb Shecter as a module in the wxPython library, named wx.lib.pubsub, sometime around y2k. At that time, pubsub had one responsiblity: to allow for messages to be sent to listeners based on a hierarchy of topics. In the Spring of 2004, I added the ability to automaticaly unregister listeners that were no longer in use outside of PyPubSub (by making the module’s Publisher use weak references to listeners). For large PyPubSub-based application, this greatly simplified listener management. I asked Robin Dunn if he would like the changes to be put in wx.lib.pubsub; he forwarded the request to Robb. And the rest is history.

Only a few minor tweaks and improvements happened for the next couple years. In 2006 when I used PyPubSub on a couple larger projects, I wished that topic trees and the topic message data could be documented. I also found that a major time waster when using pubsub at that time was debugging incorrect message data, so I started thinking of a way that I could validate message data. I also wished that I could find a design that would allow the use of tools like pylint to point out invalid topic names.

So I developed version 2 of wx.lib.pubsub in the Fall of 2006. I also created an entry in the Python Package Index as PyPubSub (http://pypi.python.org/pypi/PyPubSub) and used PyPI to hold a snapshot of my files so that even developers not using wxPython could benefit from it.

In May 2007 I decided it was time to create a project on SourceForge.net for it. It was http://sourceforge.net/projects/pubsub, so the web site was at http://pubsub.sourceforge.net. The wx.lib.pubsub was then a verbatim copy of the src folder from sf.net/projects/pubsub, as it was before PyPubSub version 2.

In 2008 someone created, unbeknownst to me, an unrelated Python project on sourceforge and named it PyPubSub. The author did not realize that mine already existed with that name in PyPI and that therefore he would have to rename his so as not to confuse users. This project came to my attention when I wanted to rename pubsub on SF.net to pypubsub to make it clear that it was python based, and to match the already one-year old entry on PyPI. In the end, the author renamed his project and sf.net/projects/pypubsub was available for my taking.

After using PyPubSub version 2 for a bit I wasn’t really happy with it, so I went back to the drawing board to support topic and message data documentation, definition and validation started. Version 3.0.0, completed some time in 2008, achieved this via keyword-based message data and topic definition providers. Version 3 also added support for tracking PyPubSub activity such as listener subscription, topic creation, and sending messages, very useful in large applications.

Version 3 of PyPubSub was not compatible with v2 or v1, so I couldn’t directly upgrade wx.lib.pubsub to it without at least supporting a deprecated v1 for a while. This led to version 3.1.0 in early 2010, which supported the v1 API via a setupv1.py configuration module that could be imported before the first import of pubsub. This was quite a challenge as there was a fair bit of commonality between PyPubSub v1 and v3, but also some significant differences. In retrospect I should not have done that because it made the code rather complex. I did a good job of it so it was easy to make fixes, but it could be a pain to troubleshoot. If I had to walk the same mile again, I would just have two separate implementations, with an easy way to import one or the other (e.g. pubsub vs pubsub3 package package folders).

Not much happened between early 2010 and first half of 2013, except for a minor release 3.1.2 in November 2011: the code was stable and did its job nicely so no major changes needed. Also in that period I didn’t develop or maintain any event-based Python application so I didn’t have any reason to update PyPubSub. I did accumulate about a dozen tickets on SF.net involving minor bugs or patches contributed by users in that period.

The overhaul of wxPython ‘Phoenix’ in 2013 was the perfect opportunity to make pubsub version 3 API the default, and to make version 1 API accessible only on demand (via the setuparg1.py configuration module). I also removed all the code that was there just to support the old version 1 API, leaving just a version 3 API with two message protocols available. I took the opportunity to address the dozen tickets during the summer of 2013, and to improve the docs.

In early 2016 I started work to remove the deprecated code and support only the original messaging protocol that I had designed in 3.0. With two busy kids, it is not easy to find the time to do this, so it took me till October 2016 for me to get my act together and finally release v4: a nice simple design with no import magic needed, no configuration, no complicated docs to explain the mulitple APIs, use of wheels instead of eggs, use of annotations, etc.

Roadmap

List of things I would like to add to PyPubSub:

  • complete implementation of multi-threading helper class, no change required to PyPubSub, rather just utility class to help user (pseudo-code already in src/contrib)

  • figure out a good way to prevent wrapped listener subscriptions from being DOA (PyPubSub only keeps weak reference to listener, so if listener subscribe like pub.subscribe( wrapper(yourListener) ) then listener will be unsubscribed as soon as subscribe returns; you need refListener = wrapper(yourListener); pub.subscribe(refListener))

  • finish the src/contrib/monitor implementation to monitor PyPubSub messages, or some way of monitoring message sending

If anyone is interested in helping, please post on the dev forum.

The following is no longer on list of things to do:

  • support pubsub over UDP and TCP sockets: mqtt does this! PyPubSub and mqtt are complementary: PyPubSub for messaging between application components within one Python interpreter; mqtt for messaging between compoonents on a network.