[Purdue-pm] On The Proper Use Of Config Files

[Dave Jacoby]

The Use Perl journal post from Alias yesterday brought up a question I 
have had for a while. I've been thinking about configuration files.

I have of course rolled my own more than once. I have also read and 
parsed variables out of the ENV variables for web apps before I 
discovered CGI, and I would occasionally try to get my own options 
before Conway pointed me to Getopt::Long. With my most recent attempt, I 
realized that I could include abstract code into my config, which seemed 
to be a security hole in the making.

So, went to the books.

The Perl Cookbook 2nd Ed (8:16) suggests you roll your own.

http://proquest.safaribooksonline.com/0596003137/perlckbk2-CHP-8-SECT-16

Perl Best Practices (19:3) suggests you use a module.

http://proquest.safaribooksonline.com/0596001738/perlbp-CHP-19

(We're all Purdue people, so if we're at work [or can tunnel] we should 
all have access. And you should have PBP anyway.)

My instinct is to follow Conway here. I have some Alias-inspired 
Config::Std-using code on the Wiki.
(http://pm.purdue.org/Wiki/wiki.pl/DaveJacobyDiary)

But to what extent is the test I do necessary? To what extent is it 
sufficient? More for following PBP than the other way. If I can't trust 
CPAN modules, how can I use CPAN modules to help me test and trust CPAN 
modules?

-- 
Dave Jacoby                         Address: WSLR S049
Purdue Genomics Core                Mail:    jacoby at purdue.edu
                                     Jabber:  jacoby at jabber.org
                                     Phone:   hah!