Volume Number: 16 (2000)
Issue Number: 4
Column Tag: MacTech Online

Of Perl and Salsa

by Jeff Clites (online@mactech.com)

If computer languages were edible, Perl would be cilantro: some people love it and some people hate it, but no one who's tried it walks away indifferent. Many Macintosh programmer may barely have heard of it, but it's a staple of the Unix world. It's also known varyingly as the Swiss army knife of the internet and the duct tape of the internet, but either way it's a piece of hardware that should be in every programmer's tool box, whether you use it frequently or infrequently. So what's so special about Perl, and why should you care? I'm glad you asked.

First, here's the clinical description: Perl is a cross-platform interpreted language, with implementations available for almost every conceivable platform (including the Macintosh and Windows). For most Unix-based systems, including Linux and Mac OS X Server, the standard distribution "just works". It's interpreted, in that Perl programs (or scripts, as they are usually called) run directly from their source code, but actually they are processed into an internal byte-code form at the start of their execution, so while running they exist in a semi-compiled form in memory, much like Java. The upshot of this is that Perl programs tend to be fast-much slower than C programs in general, but faster then other interpreted languages (such as Python and Tcl) and often faster than Java. (The Perl interpreter itself is written in C, and its speed stems in part from careful optimization, which has evolved over time.) You'll also hear Perl and its brethren referred to as scripting languages, but they are used quite differently than AppleScript and Frontier: AppleScript is mostly used to drive other applications, whereas Perl scripts tend to do most of work on their own, like traditional programming languages. It's not a hard-and-fast division, but the two types of languages do tend to be used differently.

MacPerl, the Macintosh port of Perl, has been around for quite a while, and is quite good. It's main drawbacks are that new releases always lag behind the standard version by a significant margin, and despite the magic its porter has achieved, there are still some features which are unavailable due to the nature of the operating system-for instance, you can't fork under the Mac OS (or under Windows for that matter). The good news is that with the advent of Mac OS X and its BSD compatibility layer, you can run the standard distribution-no port necessary. The current word on the street is that MacPerl will continue to evolve as a separate package for the near future, but over time I expect that it will contract to a set of modules on top of the standard distribution. This will certainly be less work for Matthias Neeracher (the man behind the port, now working in Apple's speech group), but MacPerl has several features that I hope will survive the migration, namely a nice editor and documentation viewer, and modules allowing access to a large portion of the classic Mac API.

The MacPerl Pages
http://www.macperl.com/

So, what is Perl like? To indulge in another analogy, if C and C++ are like Latin, then Perl is definitely English. It's quirky and flexible and expressive-you can always rephrase something so that the emphasis is different but the literal meaning is the same. It can be difficult to learn in a sense, but it's actually easy to get started, and once you become comfortable with the language you can write incredibly compact code. (As an extreme example, there is a two-line version of the RSA encryption algorithm.) Perl's flexible syntax is probably its most salient feature. As a quick example, like most languages Perl has an "if", but it also has an "unless", and additionally you can switch the order of the "if" and "then" parts, much as you can in English. So, you can say, "if( ! something() ) { doWhatever(); }", just as you would in C, but you can also phrase this "doWhatever() unless something();". It's just a tiny example, but it does give you a feel for they way Perl allows you to structure your code to focus attention on it's function-in this case, it lets you emphasize the action rather than the condition. It's convenient and expressive at the same time. (Perl is also the only language which has given me the odd experience of finding source code funny-not because it did anything funny or because it was poorly written, but rather because its syntax choices were a commentary on what the program was doing. I know that sounds strange, but I'll spare you the actual code.)

As another quick example of Perl's syntactic flexibility, there are at least five different ways to quote strings (and that's a conservative count). This is handy if you need a literal with embedded quotation marks or line breaks, but you want to avoid filling it with escape characters and rendering it illegible in your source code. So, rather than writing

	"\"foo\" is a made-up word."

you can write

	q{"foo" is a made-up word.}

This flexibility comes from Perl's emphasis on the practical over the merely elegant, and its terseness lets you accomplish small tasks with small code, and quickly build tools or just one-time scripts to make your work easier. Its creator, Larry Wall, is the king of the insightful quote, and my favorite line from him underscores this point: "A perl script is correct if it gets the job done before your boss fires you." For further examples of Perl's syntax, as well as answers to common criticisms of Perl, check out the "Ten Perl Myths" article on Perl.com.

Beginning Perl : Ten Perl Myths
http://www.perl.com/pub/2000/01/10PerlMyths.html

This last example brings us to Perl's most oft-mentioned strength, its string handling. Strings are a first-class type in Perl, and actually the language doesn't distinguish between strings and numbers at all-they're both just scalars. In fact it's rather awkward in Perl to treat a string as an array of characters, as they are represented in many other languages, but you'll rarely need to do so. The language's text-handling features reflect its original application of processing, analyzing, and reporting on text-based data, and it's regular expression "sub-language" provides you with the richest set of pattern-matching tools you'll find anywhere.

Also, contrary to what you may have heard, Perl does allow you to do object-oriented programming, complete with inheritance, polymorphism, and typing. But it doesn't force you to program this way, and you can ignore these feature if you just need a quick two-line program to rename a bunch of files. For those with an Objective C background, you'll be happy to hear that Perl has the functionality of categories, although I didn't recognize their power as a design tool until I encountered Objective C. (For those unfamiliar with Objective C, categories are a language feature which allows you to add methods to a class without modifying its source code, or even having access to it.)

I would be remiss if I didn't mention one of Perl's most attractive features, which isn't actually part of the language itself: the Comprehensive Perl Archive Network, or CPAN for short. It's a mirrored archive of user-contributed code, with modules for everything from FTP and SSL to parsing HTML, XML, and SGML. (Modules are the unit of reuse in Perl, and correspond to libraries or classes in other languages.) It's a reflection of the Perl user community, which is productive and helpful and opinionated at the same time, much like the Macintosh community. The value of this code base can't be overemphasized. At the same time, I can't think of any technological reason why there is nothing even close to it for C, C++, Objective C, or Java, so I'm left with the feeling that the language itself has fostered this culture of sharing. (Of course, other languages have archive sites, but the CPAN tends to be well-organized and there is, for example, one Net::FTP module rather than 10 different implementations. For the record, I believe that Python's archive also tends to be well-managed.)

CPAN: Comprehensive Perl Archive Network
http://www.cpan.org/

I hope to return to Perl in the coming months, and cover some specific applications of the language in more detail. Until the, I'll leave you with pointers to a few books designed to help you learn Perl. Also, be sure to check out the Perl.com web site, which has a wealth of resources and HTML versions of all of Perl's documentation, much of which is included with the distributions themselves. The information there is excellent but can be a bit overwhelming for someone coming to the language for the first time. For printed material, the book which jumps to mind first is MacPerl: Power and Ease, by Vicki Brown and Chris Nandor (published by Prime Time Freeware). It is indispensable if you are using the Macintosh version of Perl, but valuable even if you are not. (I hope that the authors bring us a second volume at some point.) The classic for non-Macintosh-specific Perl is Learning Perl, published by O'Reilly and Associates, who have produced a majority of the high-quality Perl titles out there. Finally, there are two new books by Manning Press which have been drawing high praise, Elements of Programming with Perl and Object Oriented Perl. The latter is not geared at beginners specifically, but takes the unique approach of treating Perl as an object-oriented language from the start.

Perl.com
http://www.perl.com/
MacPerl: Power and Ease
http://www.macperl.com/ptf_book/
The O'Reilly Perl Center
http://perl.oreilly.com/
Elements of Programming with Perl
http://www.manning.com/Johnson/
Object Oriented Perl
http://www.manning.com/Conway/

If you want to broaden your linguistic horizons further, come to the MacTech Online web pages at http://www.mactech.com/online/, where you'll find links to other fun languages.