I've been working on a smallish but vital system (written in Perl, or course) for the past few months. The other day a question came down from VP to Director to me: "Is PERL the most optimal way for us to do that?"
now, that's about the most loaded question I can think of - probably rooted in the old Perl == CGI idea, washed down with an unhealthy dose of java kool-aid.
so, after deciding that being able to handle 4000 transactions a minute (when we only get 3000 a day) probably wouldn't be a sufficient answer, this morning I took a moment to analyze my smallish system in ways that would matter to me if I were an (unbiased) VP:
- 14 OO classes (with inheritance, etc)
- 7645 lines of (well commented) code
- 3687 lines of documentation (real pod, not including above comments)
- 654 tests (that have saved me more than once)
I'd wager that the test suite alone makes it the most optimal piece of software in the company, but I don't think that's what he meant :)
Re:"Optimal" means many things
geoff on 2002-12-20T00:24:42
what I was trying to do in my own (brief) analysis is defend exactly the three points you bring up: yes, the software is supportable (well documented); yes, others will be able to maintain it (with a test suite, at least they can't break anything that currently works); yes it's a serious language (it can take advantage of OO like everyone else).
I actually didn't expect that he meant performance:)
--Nat
Re:Don't forget
geoff on 2002-12-20T00:34:49
actually, I suspect that he's leaning either toward java (the company standard) or something more abstract, like "XML" - both of which fit the nature of the project. What he will likely be shocked to find is that I've been advocating (and capable of supporting) the use of direct SOAP calls to my API for a while (as opposed to the daemon approach it uses now -SOAP::Lite++
). I keep hoping somebody will take me up on it so I can giggle as the Java people try and keep up:)