The discussion on this thread has made me recollect some of my own experiences. I had been hacking Perl for 3 years before seeing the light and discovering use strict;. There are plenty of places on this site which extoll the virtues of strict.pm, so I will not be going into details.

My initial experience was Waah! sooo many errors! But then I looked in detail at what the code was doing (all the code was 100% written by me - no cargo deliveries here :-), I realised that 99% of the errors could be removed just by inserting the keyword my. There was no need for any of the offending variables to be global.

Also, I had mastered subs' argument passing and returning quite early on (thanks to the Camel 2nd edition and Sriram Srivanam's Advanced Perl Programming), which meant I never needed to pass in values to subs via global variables. Having a Comp sci and C programming background, I instinctively felt that global variables were ugly.

I was thus able to iterate round the loop:

Edit source.pl
perl -c source.pl
understand errors 

The process, uncovered some bugs which I had no idea existed. For example, the newbie trap

my $foo, $bar;
[download]

Only $foo here is lexical. I was wondering at the time of writing the code originally, why $bar was retaining values between iterations.

The remainder of the strict errors fell into 2 categories:

• Variables in subs that genuinely needed to retain values between calls. I learned the trick of using 'my' variables in an outer block to the sub, and learned all about closures in the process.
• Symbolic references. Many of these were rewritten to use data structures. Other instances the code was genuinely doing meta-symbolic generic stuff, so I embedded the critical code in blocks with no strict 'refs';

Having got a clean compile on strict, I then did the same iteration loop with -w.

I recommend this approach - it's good for the soul. Also, I find that a consequence is that any new code I write, my fingers go on auto-pilot with use strict; even when coding a noddy example.