Should FileExchange submissions still support Matlab < 7.0?

Matlab 7.0 was published in 2004 and for basic programming the leap from 6.5 was larger than from 7.0 to 8.0. As an indicator, if Matlab 6.5 is still in use, we could examine the threads in this forum concerning older releases. Most of them have been written many years ago, but this is not a representativ survey. I'm not sure if TMW can offer hard numbers of users of historical releases, so I ask for your opinions:

Is it still useful to support Matlab 6.5 (R13) in functions published in the FileExchange?

[EDITED, clarified that the question concerns the FEX, not the forum]

Answer 1

[Andreas Goser]

My recommendation to FEX authors is to test submissions in each release and update the compatibility information.

For older releases, I would make sure, I reach at least 80% of the user base that is like the last 4 releases for MATLAB based applications and maybe 6 releases for Simulink based applications. But this is my personal viewpoint.

[Jan]

As a member of the support team, your estimation is valuable. When about 80% uses a release > 2013a, supporting < 2011a in FEX submissions is not efficient.

Thanks!

Answer 2

[Andreas Goser]

Hello Jan, your primary point seems to be whether information for old releases should be maintaind on MATLAB Answers or not (You mention FileExchange, but your example is MATLAB Answers).

Actually, there was a "clean up" a year ago based on a couple of factors and one factor was the number of times a certain page was accessed. So in your example above, you can be sure that there were users who searched for this and found it!

As you are following MATLAB communities for a longer time, you may have recognized that the wording has changed over time with MathWorks Support provided answers. Being specifc about a release can be very important, but it can be also confusing. So today your example above would probably just be "Is MATLAB supported on IRIX?" or "What is the MATLAB release compatibility on IRIX?".

[Image Analyst]

So when we see new or newly-edited, discussions that still specifically mention a particular release, such as this one http://www.mathworks.com/matlabcentral/answers/164988-how-do-i-incorporate-a-constraint-into-my-portfoliomad-optimization-in-matlab-r2014a we can assume that it is because there's something particular about that exact release , and not any other releases, correct?

[Jan]

@Andreas: I took the examples from answers as an indicator, because there is no feedback mechanism in the FileExchange concerning the Matlab versions. My actual problem concerns my submissions in the FileExchange. I've edited the question for a clarification.

But your answer is very interesting. Obviously it would be helpful to see how many times a thread was opened.

[Andreas Goser]

OK, answering about FEX seperately and here the Image Analyst's comment. In gereral, the information should be as specific as possible, but on the other hand you also want to avoid waste.

The example is about the "PortfolioMAD" article you linked from the perspective of a support engineer. When something is reported the first time (that was here the case) and do not know about imapct on other release, you be specifc about the release. Whether something is relevant to different release or not may be obvious or needs research. The question on whether you invest the time to do the research depends on your assessment on how often users run into it. E.g. when you use and article with this text and find it was needed a couple of more times, you may want to invest and research and edit.