Post Syndicated from Bradley M. Kuhn original http://ebb.org/bkuhn/blog/2012/12/18/perl-cobol.html
In 1991, I’d just gotten my first real programming job for two reasons:
nepotism, and a willingness to write code for $12/hour. I was working
as a contractor to a blood testing laboratory, where the main
development job was writing custom software to handle, process, and do
statistical calculations on blood testing results, primarily for
paternity testing.
My father had been a software developer since the early 1970s, and worked
as a contractor at this blood lab since the late 1970s. As the calendar
had marched toward the early 1990s, technology cruft had collected. The
old TI mainframe, once the primary computer, now only had one job left:
statistical calculation for paternity testing, written in TI’s Pascal.
Slowly but surely, the other software had been rewritten and moved to an
AT&T 3B2/600 running Unix System VR3.2.3. That latter machine was the
first access I had to a real computer, and certainly the first time I had
access to Usenet. This changed my life.
Ironically, even on that 3B2, the accounting system software was written
in COBOL. This seemed like “more cruft” to me, but fortunately
there was a third-party vendor who handled that software, so I didn’t have
to program in COBOL.
I had the good fortune, actually, to help with the interesting problems,
which included grokking data from a blood testing machine that dumped a
bunch of data in some weird reporting format onto its RS-232 port at the
end of every testing cycle. We had to pull the data of that RS-232
interface and load the data in the database. Perl, since it treated
regular expressions as first-class citizens, and had all the Unix block
device fundamentals baked in as native (for the RS-232 I/O), was the
obvious choice.
After that project, I was intrigued by this programming language that had
made the job so easy. My father gave me a copy of the Camel book —
which was, at that point, almost hot off the presses. I read it over a
weekend and I decided that I didn’t really want to program in any other
language again. Perl was just 4 years old then; it was a young language
— Perl 4 had just been released. I started trying to embed Perl into
our database system, but it wasn’t designed for embedding into other
systems as a scripting language. So, I ended up using Tcl instead for the big
project of rewriting the statical calculation software to replace the TI
mainframe. After a year or two writing tens of thousands of lines of Tcl,
I was even more convinced that I’d rather be writing in Perl. When
Perl 5 was released, I switched back to Perl and never really looked
back.
Perl ultimately became my first Free Software community. I lurked on
perl5-porters for years, almost always a bit too timid to post, or ever
send in a patch. But, as I finished my college degree and went to graduate
school, I focused my
thesis work on Perl and virtual machines. I went to the Perl
conference every year. I was even in the room for the perl5-porters
meeting the day
after Jon
Orwant’s staged tantrum
, which was the catalyst for the Perl 6
effort. I wrote more than a few RFC’s during
the Perl 6 specification
process. And, to this day, even though I’ve since
done plenty of Python
development, too, when I need to program to do something, I open an Emacs
buffer and start typing #!/usr/bin/perl.
Meanwhile, I never did learn COBOL. But, I was amazed to hear that
multiple folks who graduated with me eventually got jobs at a health
insurance company. The company trained them in COBOL, so that they could
maintain COBOL systems all day. Everyone once in a while, I idly search a
job site for COBOL. Today, that search is returning 2,338 open jobs. Most
developers never hear about it, of course. It’s far from the exciting new
technology, but it’s there, it’s needed and it’s obviously useful to
someone. Indeed, the COBOL standard was just updated 10 years ago, in
2002!
I notice these days, though, that when I mentioned having done a lot of
Perl development in my life, the average Javascript, Python, or Haskell
developer looks at me like I looked at my dad when he told me that
accounting system was written in COBOL. I’d bet they’d have my same
sigh of relief when told that “someone else” maintains that
code and they won’t have to bother with it.
Yet, I still know people heavily immersed in the Perl community. Indeed,
there is a very active Perl community out there, just like there’s an
active COBOL community. I’m not active in Perl like I once was, but it’s a
community of people, who write new code and maintain old code in Perl, and
that has value. More importantly, though, (and unlike COBOL), Perl was
born on Usenet, and was released as Free Software from the day of its first
release, twenty-five years ago today. Perl was born as part of Free
Software culture, and it lives on.
So, I get it now. I once scoffed at the idea that anyone would write
in COBOL anymore, as if the average COBOL programmer was some sort of
second-class technology citizen. COBOL programmers in 1991, and even
today, are surely good programmers — doing useful things for their
jobs. The same is true of Perl these days: maybe Perl is finally
getting a bit old fashioned — but there are good developers, still
doing useful things with Perl. Perl is becoming Free Software’s COBOL:
an aging language that still has value.
Perl turns 25
years old today. COBOL was 25 years old in 1984, right at the time
when I first started programming. To those young people who start
programming today: I hope you’ll learn from my mistake. Don’t scoff at the
Perl programmers. 25 years from now, you may regret scoffing at them as
much as I regret scoffing at the COBOL developers. Programmers are
programmers; don’t judge them because you don’t like their favorite
language.
Update (2013-04-12):
I posted
a comment on Allison Randal’s blog about similar issues of Perl’s
popularity.
