I think we should ignore the man-hour calculations which are a bit of an eye-roller.
Really? And why should that be? Let's take a look at the hours I've spent on Excel::Template.
- I wrote E::T as a fork of PDF::Template, which I had taken over from Dave Ferrance. I would estimate he had spent between 100 and 250 hours on the 0.05 version. My initial release took 4 weeks to test, which is 160 hours, plus about 40 hours of design help and testing from a coworker. (200-450 hours)
- The actual forking of E::T took about 4 days, resulting in an initial release cost of 30 hours. (230-480 hours)
- I've released 24 updates to E::T. The minimum amount of work to release an update to a CPAN module is 2 hours. I generally put in about 5-20 hours per release, averaging about 10 hours per release. So, that's 12 hours per release, on average. So, 24 * 12 is 288 hours. (518-768 hours)
So, let's call it 600 hours of work put into Excel::Template. At a measly $25/hr (which is what sloccount appears to use), that's $15,000. For one distribution. CPAN has over 7000 distros. If we assume E::T has an low-to-average cost (which I would say is about fair, given the Acme namespace vs. DBI or CGI), then that's a
minimum of $105 million dollars. Hmm ...
My criteria for good software:
- Does it work?
- Can someone else come in, make a change, and be reasonably certain no bugs were introduced?