[Israel.pm] Perl version migration

Gabor Szabo gabor at pti.co.il
Sun May 2 07:38:00 PDT 2004

On Sun, 2 May 2004, Jason Elbaum wrote:

> A good chunk of our code is written by engineers, not programmers.
> Software testing is not their field. And no one's going to start playing
> with old script systems that worked perfectly well in the past.

Most of the "perl programmers" are either engineers, system
administrators, biologists, or people without a degree. Such as
MJD or Michael Schwern the guy behind most of the automated testing
modules in Perl.

> No, not at all. Why should disaster occur if the environment hasn't
> changed since the scripts last ran?

If your OS never changes, the requirements of this scirpt don't
change, the input never changes and the input is always tested[1] in case
a new user does not know exactly what to enter then probably the only
case is the changing of the moon phase.

[1] But you said most of the scripts are written by engineers who don't
understand in programming so I can assume there might be a few cases they
don't check all the input values.

Really, how many of those old script need changing once in a while because
of changing requirements ? How do you do the change ? Copy the script and
do the change on the new script ?

> Deliberately changing the Perl
> version, on the other hand, is an invitation for disaster. Especially
> when certain scripts run only occasionally - if they fail, who's going
> to know if they fail due to the version of Perl or the phase of the moon?

This is the reason other companies move the scripts one by one and don't
change the Perl version under the untested scripts.

> See  http://www.freescale.com/ for more details.
Actually the few lines you wrote were more informative than that web site


More information about the Perl mailing list