Thread Subject: Do you upgrade every 6 months?

I am curious if there is a general usage model of when people upgrade their version of MATLAB. I have software maintenance so I get the upgrades, but I do not tend to use them. I install the newest release of MATLAB that is available when I get a new computer, but I do not upgrade MATLAB unless the computer crashes. Since I have a number of different computers, each one has a different version of MATLAB (as early as 2009a).

I do tend to look at the release notes hoping for something exciting that would motivate me to upgrade. Sometimes I find things that are scary and seem likely to cause backwards/forwards compatibility issues (e.g., changes to the RNG and the ~ notation). The last time I felt motivated to upgrade was 2008a(?) and the changes to OOP.

On May 9, 9:04 pm, "Daniel " <daniel_s...@yahoo.com> wrote:
> I am curious if there is a general usage model of when people upgrade their version of MATLAB. I have software maintenance so I get the upgrades, but I do not tend to use them. I install the newest release of MATLAB that is available when I get a new computer, but I do not upgrade MATLAB unless the computer crashes. Since I have a number of different computers,  each one has a different version of MATLAB (as early as 2009a).
>
> I do tend to look at the release notes hoping for something exciting that would motivate me to upgrade. Sometimes I find things that are scary and seem likely to cause backwards/forwards compatibility issues (e.g., changes to the RNG and the ~ notation). The last time I felt motivated to upgrade was 2008a(?) and the changes to OOP.

I agree that upgrades can look scary.
Indeed, I opted out of upgrades when I found that way back then, when
I upgraded from 2006a to 2006b, several programs that worked in 2006a
no longer worked. So, to get a miniscule enhancement in Matlab every
6 months, I had to spend hours re-coding .m files.

So, I saved myself some money and gave up on upgrades.
I'm still using 2006a and see no reason to upgrade, even now.

TideMan <mulgor@gmail.com> wrote in message <ab7fab29-4b74-453f-ba00-e8b1293256f4@z15g2000prn.googlegroups.com>...
> On May 9, 9:04 pm, "Daniel " <daniel_s...@yahoo.com> wrote:
> > I am curious if there is a general usage model of when people upgrade their version of MATLAB. I have software maintenance so I get the upgrades, but I do not tend to use them. I install the newest release of MATLAB that is available when I get a new computer, but I do not upgrade MATLAB unless the computer crashes. Since I have a number of different computers,  each one has a different version of MATLAB (as early as 2009a).
> >
> > I do tend to look at the release notes hoping for something exciting that would motivate me to upgrade. Sometimes I find things that are scary and seem likely to cause backwards/forwards compatibility issues (e.g., changes to the RNG and the ~ notation). The last time I felt motivated to upgrade was 2008a(?) and the changes to OOP.
>
> I agree that upgrades can look scary.
> Indeed, I opted out of upgrades when I found that way back then, when
> I upgraded from 2006a to 2006b, several programs that worked in 2006a
> no longer worked. So, to get a miniscule enhancement in Matlab every
> 6 months, I had to spend hours re-coding .m files.
>
> So, I saved myself some money and gave up on upgrades.
> I'm still using 2006a and see no reason to upgrade, even now.

I do, although I need to maintain code in a flexible state
so that it can be used by a variety of older releases,
currently back to about 2006.

It takes me 15 minutes (roughly) to upgrade with the
reasonably fast connection that I have. Most of that
time is spent in resetting the search path.

With 8 cores on my machine, the newer versions of
matlab are actually making good use of them often
enough to save me time.

That alone is well worth it. Why have a machine with a
gazillion cores, only to let them stand idle most of the
time?

As far as updating code goes, I have 20+ years old code
that is still being used by past clients, with essentially no
changes made. The rest of it has been gradually kept up
to date at little cost in time. When something starts
sending out messages that a given call will be made
obsolete in the future, that is the logical time to do the
repair, not when it suddenly stops working.

John

On May 10, 12:03 am, "John D'Errico" <woodch...@rochester.rr.com>
wrote:
> TideMan <mul...@gmail.com> wrote in message <ab7fab29-4b74-453f-ba00-e8b129325...@z15g2000prn.googlegroups.com>...
> > On May 9, 9:04 pm, "Daniel " <daniel_s...@yahoo.com> wrote:
> > > I am curious if there is a general usage model of when people upgrade their version of MATLAB. I have software maintenance so I get the upgrades, but I do not tend to use them. I install the newest release of MATLAB that is available when I get a new computer, but I do not upgrade MATLAB unless the computer crashes. Since I have a number of different computers,  each one has a different version of MATLAB (as early as 2009a).
>
> > > I do tend to look at the release notes hoping for something exciting that would motivate me to upgrade. Sometimes I find things that are scary and seem likely to cause backwards/forwards compatibility issues (e.g., changes to the RNG and the ~ notation). The last time I felt motivated to upgrade was 2008a(?) and the changes to OOP.
>
> > I agree that upgrades can look scary.
> > Indeed, I opted out of upgrades when I found that way back then, when
> > I upgraded from 2006a to 2006b, several programs that worked in 2006a
> > no longer worked.  So, to get a miniscule enhancement in Matlab every
> > 6 months, I had to spend hours re-coding .m files.
>
> > So, I saved myself some money and gave up on upgrades.
> > I'm still using 2006a and see no reason to upgrade, even now.
>
> I do, although I need to maintain code in a flexible state
> so that it can be used by a variety of older releases,
> currently back to about 2006.
>
> It takes me 15 minutes (roughly) to upgrade with the
> reasonably fast connection that I have. Most of that
> time is spent in resetting the search path.
>
> With 8 cores on my machine, the newer versions of
> matlab are actually making good use of them often
> enough to save me time.
>
> That alone is well worth it. Why have a machine with a
> gazillion cores, only to let them stand idle most of the
> time?
>
> As far as updating code goes, I have 20+ years old code
> that is still being used by past clients, with essentially no
> changes made. The rest of it has been gradually kept up
> to date at little cost in time. When something starts
> sending out messages that a given call will be made
> obsolete in the future, that is the logical time to do the
> repair, not when it suddenly stops working.
>
> John

John

It's usually more subtle than obsolete commands.
In 2006a, when you execute:
matlab -r myfile
it opens a window executes the command and closes Matlab.
In 2006b, the Matlab session remains open - some TMW bright spark
decided to change it without telling anyone.
I have dozens of Matlab scripts operating on an hourly schedule.
The morning after I installed 2006b, there were about 400 Matlab
sessions open.
I had to go through and change each hourly script to:
matlab -r myfile;exit

Why should I have to do that?
Especially when I pay good money to get an upgrade that is a miniscule
improvement otherwise.

"John D'Errico" <woodchips@rochester.rr.com> wrote in message <iq8l5o$qb2$1@newscl01ah.mathworks.com>...
> With 8 cores on my machine, the newer versions of
> matlab are actually making good use of them often
> enough to save me time.
>
> That alone is well worth it. Why have a machine with a
> gazillion cores, only to let them stand idle most of the
> time?
>
Yes, you are right. I guess I look at the updates as having two parts. Things that break backwards compatibility (changes to the MATLAB langauge) and things that improve the MATLAB experience. Support for 64-bit OS's and muilt-core processors, and improvements to mlint, the editor and desktop improvements have enhanced MATLAB, but have not changed the language. Changing the random number generator and the way classes are defined are core changes to the MATLAB language in my mind. I thnk it is these changes that frustrate me. I wish they were lumped together and released on a longer timeframe (maybe 2 years).