Problems with system() call.

This is breaking our production runs, thanks for the help.

We have R2013b executing on Ubuntu 14.04:

Linux kashmir 3.13.0-32-generic #57~precise1-Ubuntu SMP Tue Jul 15 03:51:20 UTC 2014 x86_64 x86_64 x86_64 GNU/Linux

Our job chokes on the empty 'user' variable:

>> [status, user] = system( 'echo -n $DB_FULL_USER' )

status =

       0

user =

     ''

Ooops, next time we make the call we get an erroneous doublet:

>> [status, user] = system( 'echo -n $DB_FULL_USER' )

status =

       0

user =

matlabmatlab

Then on the next call we go back to normal for a 'while':

>> [status, user] = system( 'echo -n $DB_FULL_USER' )

status =

       0

user =

matlab

The painful thing is that this is years old code, never broke before. AFAIK it's intermittent. Very strange, and I don't know what to do about it other than to loop on system(), but that seems crazy. Anybody have any idea what's happening?

THANK YOU

Clark

Answer 1

[Clark]

This post appears to be related:

http://stackoverflow.com/questions/21120296/matlab-getenv-unable-to-find-some-linux-environment-variables.

Here we find that:

map = java.lang.System.getenv();

seems to be the means used by getenv() to attain its results.

whereas:

[ status, blahVal ] = system( 'echo -n $BLAH' )

could not "possibly" be using java. We're going to try switching our usage to getenv(). A few more days and we'll know if it's any better.