Requesting feedback from the Staab/Snyder presentation

[Daklu]

Thanks to everyone who took the time to attend our presentation dry run.  (And a special shout out to Yair for getting up at 3 am to attend.)

We ran way over the 50 minutes we are allotted and need to be ruthless with the scissors.  My part of the presentation took nearly 30 minutes--my budget is 15-20 minutes.  I certainly need to rehearse more and that will help reduce the time, but I don't think I can hit my target just by talking faster.    

My goal is to help people understand actor-oriented design on a deeper level than "this is what the code looks like."  Once the whys are understood the hows (the implementation) are almost an afterthought. From my part of the presentation, what did you find most helpful for understanding the essence of AoD?  What did you find least helpful?  Please also include a brief description of your experience with AoD and the Actor Framework.

Thanks!

[mike_nrao]

Great presentation!  Thanks guys.

My comments on Dave Snyders portion:

I think the most important aspects (you emphasized them appropriately) are

. AoD concepts are separable from OOD

. Messages are requests not commands

. the MHL is not a job queue

. not everything has to be an Actor

If you have to ditch some slides, please don't drop the "theory" type ones at the beginning.  I think these are the most valuable.  A little more emphasis on benefits of Hierarchical topology would please me, stressing that the danger of sharing a messaging interface with many actors is akin to global variables scattered throughout code.

In Staabs portion: The NI Week audience will differ from yesterday's audience and may benefit more if you bring it down to earth with some real-world examples like: Related to error handling and/or message acknowledgment -- An actor performing a measurement procedure is waiting for an actor who's responsibility is to move a motor into position; Acknowledgement times-out or motor actor fails and reports an error; then we override default behavior to cause the measurement actor to simply skip the next step in the procedure by modifying the actor's state and not propagating the error.

These types of explanations helped me as a newbie as I tried to connect and apply concepts to the daily grind.

[pavanb]

Where can I get the presentation/video?

[Daklu]

As far as I know nobody recorded the dry run we did for architects forum.  However, you can download the video of my part of the presentation at NI Week 2013.  Follow the link from this post to get instructions.  Once you logon to the ftp server, the video can be found at:

/2013 NIWeek/Presentations_avi/David Snyder - Daklu_Principles of Actor Oriented Design.avi

[David_Staab]

You can also get a copy of the slides here if you want my content.