A Call for Better LabVIEW Design Patterns

[PaulLotz]

OK, I finished the first version.  It's a bit rough around the edges in places, but I wanted to get it out.  We posted it on our site: http://www.lowell.edu/dct/content/publications/OOMessagingCommandPatternStatePattern-1.pdf

In addition to messaging with objects it gives examples of the Command and State Patterns, with a specific example that writes configuration information to XML files.  Hopefully it will help someone, and I hope it sparks conversation to refine the approaches.

[crelf]

Nice job Paul - that's a really good start!

I have been talking to several LabVIEW architects of late, and we're thinking of putting together a book - a go-to resource for design patterns.  I know there are several resources already, but nothing that really covers all the standard patterns in sufficient detail (including theory behind them as well as examples), or the nouse to choose which pattern is most appropriate.  We think it would have enough of an audience to put the work into it, what do you guys think?  Also, I had thought of it as a collaboration (an author could be respoinsible for a pattern or two) - is anyone here interested in contributing?

[Nate_Moehring]

This is a great idea Christopher, thanks for offering it up to the

community! I'd love to be involved. I don't consider myself an expert in

SE design patterns, but I love to learn about design patterns, and I bet I

can become an expert on 1 pattern.

One thing I can offer up to the community is a presentation I put together

for the Phoenix Area Advanced LabVIEW User Group. In fact Paul's

presentation will also be shared at in this arena in September, which I'm

very excited about. My presentation

(http://decibel.ni.com/content/docs/DOC-11524) tries to illustrate the

examples given by Mercer's Common OO Design Patterns document. It's not

the prettiest presentation the world, but the process of learning about the

patterns enough to teach them to others definitely strengthened my

understanding and I think it might help others too.

I'm a little concerned about a printed book though. Do we think the

features of LabVOOP have stabilized enough that a book of design patterns

would stay current for 3-5 years? If such a book were to have been printed

before LabVOOP, DVRs or the IPE, the book would now be a waste of paper.

Nate Moehring

Senior Software Engineer

Certified LabVIEW Architect

Orbital Sciences Corporation

480-355-7854

"The time is always right to do what is right." - MLK Jr.

[PaulLotz]

Christopher,

I think assembling such a resource would be a great idea!  I do think that, much as the original Design Patterns book was a collaborative effort, the LabVIEW-based version would benefit from reviewing the best implementations so far to select the best approaches and to understand the trades of different options, an effort which will require some discussion among the authors.  That to me is the main point of having design patterns in any case, and the resulting implementation examples could be quite fruitful to the community.

Paul

[Daklu]

Chris,

The book you describe is certainly needed for advanced LV developers.  Whether there's enough advanced LV developers to interest a publisher in funding might be debatable, though I think it would create its own market to some extent.  I would also be concerned about the information in the book being relevant and correct over the next several years--not so much because of possible changes to LabVOOP (though as Nate pointed out that is a concern,) but because I'm not sure LabVOOP design patterns themselves have had enough time to mature.

It is obvious this is a fast moving area of research.  Personally I'd really like to see more open dialog and code sharing between everyone working on these things.  Some sort of semi-public source code repository with formal or informal collaborative teams made up of NI engineers and advanced LV developers.  (Perhaps even restricting it to CLA's if necessary.)  Getting more eyes on a potential pattern is only going to improve it faster.  Seems to me that's to everyone's benefit.

-Dave

(Edit - Yes, I'd be interested in contributing.)

[Ben]

No way management will allow contributing to a book.

I agree that an open discusion of LV implementation of design patterns would be a nice resource for others while also letting us levearge the existing work on Design Patterns.

I have a three-ring binder wiht hard-copy print-outs of many of the Wikipeadia articals re:Design Patterns. I am still learning them but they are giving a good ouline to work from.

But open discusion about the patterns and how they can be implemented would be good. A careful analysis of the negative impacts of each (Factory Design Pattern - need to know all at development ???) as well as performance for each variation.

So I can't lead this game but I can question and yap.

Ben

[Daklu]

Ben wrote:

I have a three-ring binder wiht hard-copy print-outs of many of the Wikipeadia articals re:Design Patterns. I am still learning them but they are giving a good ouline to work from.

Wikipedia articles on design patterns are correct, but IMO they are terrible teaching material.  Spend the $30 for Head First Design Patterns.  It teaches design patterns using fictional customers wanting applications with requirements that change over time.  It explains why the existing design can't satisfy the new requirements very well, and then shows how to refactor the design using one of the patterns.

I promise (well... I'm highly confident) reading the book will jumpstart your understanding of design patterns.  Code examples are in Java, but they are simple enough to understand for anyone who is somewhat familiar with OO syntax in text languages.  (i.e.  Object.Method()  )

[Ben]

After I finish Craig Larmann's book.

Ben

[crelf]

See, that’s what we’re thinking about as far as a book is concerned: not something to completely replace Head First, but a book with enough detail on the patterns where all the example source code is in LabVIEW – something that brings together the theory of the Wikipedia articles, the more verbose descriptions and applications from Head First, source code in LabVIEW, as well as some advice on a design pattern selection tree.

I totally agree that the pattern applications based on LVOOP aren’t totally mature, and I fully expect them (as well as patterns based on other technologies) to grow in coming years, but I think that the anticipation of a timeline is a great motivator. Also, I expect that places like this online community are going to be perfect for forming the book, as well as continuing the discussion well after the first edition is printed. For that, and other reasons, I’d expect there to be several editions in the coming years. We’re talking about a dedicated LabVIEW Design Patterns resource – both a book with a CD, as well as an online community. At the moment, there’s no organized area for this, online or otherwise. I see plenty of excitement and activity in the area of design patterns, especially with the introduction of LVOOP to the masses (and, to a lesser extent, the emerging DVR concept), but we’re trying to steer and combine it into a more universally accepted and organized format. That’s one of the reasons Head First is so popular – it took what was an unorganized collection of thoughts and presented them in an easy-to-understand authority.

[Daklu]

Sign me up. 

Seeing as how you're one of the few people here who have actually been through the book writing process, are you leading the effort?