|Just another Perl shrine|
I've thinking about what's beyond the basics of Perl, and here's where everyone gets to expand my thinking and provide a reality check. What comes next? What's the "upgrade path"? Do we keep teaching them more facts about Perl until they know all of the documentation? Teach them to use the modules they need to use to get their particular job done? Or is it something else?
I just finished working on the fourth edition of Learning Perl, which should be out towards the end of summer. (I don't have details on anything yet, but I'm working with the O'Reilly marketing people to get all that stuff. And yes, Randal and Tom worked on it: I was just renamed "Dances with Wolves and Editors" ;) Teaching beginners is easy because we know where to start and everyone is pretty much in the same place. A lot of people are content to stay at that level, and that might be okay for them. What catapults people into the higher levels, though?
Besides specialized topics like CGI, LWP, Testing (chromatic and Ian Langworth are quickly approaching a completed manuscript that looks really, really good), there isn't a long vertical path, or even a contiguous one. People's interest diverge. At Stonehenge, after we teach the intermediate Alpaca course, it's either a buffet of shorter classes on specific topics or custom-designed courseware. It's also a captive, targeted audience. What about the Perl world at large? If we can't group them into a particular class because that's the class they signed up for, what do we pick out of all the things that exist as the important things to teach or write about?
Beyond Learning Perl, which we designed to cover the 80% of Perl that most people will need to know to write basic programs, there is Learning Perl References, Objects, & Modules (which we are now updating), Simon Cozen's soon-to-be-released update to Advanced Perl Programming (which is a very cool book, having just read the pre-galleys), Mark Jason Dominus's Higher-Order Perl (which he talks about in my interview with him), and Damian's upcoming Perl Best Practices (which I'm now reading in pre-proofs). There are a few other general subject books too, so my apologies to those I've neglected. I've read too many books this week. Most of these, however, focus on something to do with the Perl language (surprise!)
The more I think about this, the more I'm convincing myself (for good or ill), that the things programmers need to learn next aren't so much about Perl as things they should know how to implement in Perl, but could also do in some other language. Surely I can teach more and more Perl stuff, but at some point the student who really wants to learn more needs to have a "programming lifestyle" that supports it.
I've jotted down a couple of these lifestyle issues. They aren't revolutionary or new, and I don't have complete discussions on them. When I started this, I wanted to write the complement of How not to code, but I really think mastering Perl (or any language) goes beyond the stuff that other people can tell you: you need to do things yourself.
You need a safety net
Most of us know we should use source control and write tests, but I'm starting to think that to teach advanced concepts, people need a base camp to start their exploration. If you mess things up, you haven't completely horked your production system or corrupted your operating system. What else does the safety net comprise?
Once you have the safety net, just try things. You shouldn't have to ask questions like "What happens if I do this?" because you do it and find out. An expert is someone who has made every mistake, so you have to start making more mistakes. Or, in a foodie analogy, I heard an interview with a food writer who said "If you aren't getting food poisoning, your not being adventurous enough". To experience the really good things, sometimes you have to be bold and take risks. Maybe you get burned a couple times, but the times you don't are worth it.
You have to know more
A lot of advanced stuff is just knowing more, and that usually means knowing things that aren't immediately useful. You can't wait until you need something to learn it if you want to do fancy things since fancy things tend to be the cobbling together of many things.
I think one of the greatest enemies of this part is "Search". When you can jump right to the part you want, you often miss a lot of unrelated but interesting things that surround it. You don't get the opportunity to subconsciously remember stuff. I have a love-hate relationship with online dictionaries because I miss running my finger down a column of words and learning a word or definition I didn't intend to learn.
As part of this, I recommend to my beginner Perl students that they read all of perlvar and perlfunc, although I stress they don't have to try to remember any of it. Some day in the future, they may remember that there is a function called gethostbyname (or kill, or whatever). They may not remember exactly what it was or how it worked, but they should recall that Perl has a lot of built-ins that deal with socket stuff.
Learn other languages
In a Perl class or book, I'm not going to teach Lisp. MJD talks about Lisp in Higher-Order Perl, but not enough to distract from the Perl. Still, the book is all about using ideas from functional programming. In a recent thread, Randal talked about a particular Smalltalk book.
Other languages give you perspective on the one that you want to use. You might hate Java, or C++, or something else, but the honest programmer has to admit that smart people developed those things and had at least a couple of good ideas. Even the bad ideas simply make you appreciate your language of choice that much better.
Once you know other languages, even if you only can read them (rather than create useful things in them), you can read other books on programming to learn high-level skills and techniques which you can then apply to Perl.
Knowing that, do we really need a "Do Bar in Foo" book for every combination of task and language? You cut yourself off from mastering your favorite language if you don't pay attention to the good ideas and idioms in other languages.
Learn how to answer your own questions
The best part of my college education were the classes where the professors wouldn't answer questions. It's not that they didn't want us to ask questions: they were training us to be able to answer questions on our own. Every upper level chemistry class I had in my undergraduate days had a tough library assignment that came with no instructions other than "the answer is in the library". A couple classes were merely "this instrument needs to be rewired, and here's a bottle of mercury to build the barometer you'll need. It's due next week. Good luck." To be certain, it sucked, but after a while, I knew how to find information and answer my own questions. I didn't even have Google then.
Or do you just ask questions as Anonymous Monk? :)
Reinvent the wheel
In Advanced Perl Programming, Second Edition, Simon talks about the "rites of passage". Those are the things that the intermediate programmer has to go through to pass onto the ranks of advanced programmer, and they are all about re-writing something that is usually already done much better elsewhere: templating engine, command line argument parser, and so on.
There's some inflection point in the learning path where all that good advice we give to beginners is bad advice. Stuff like "use a module" and "just use a regex" get in the way of "kata", where the intermediate programmer needs a topic against which he tries new techniques and approaches.
Back in the day when I was answering a lot of questions on comp.lang.perl.misc, instead of giving just one code sample, I'd give several. I tried to get each answer to do the job with a different feature. One might use a regex while another one uses index().
Not only that, but it's in re-inventing something that you discover just how difficult the problem is, and you appreciate design decisions and limitations in what already exists. It's almost like you have to bang your head against the wall or touch the hot frying pan to learn that those things hurt. It's empirical learning: there is no knowledge without experience.
You don't have to use the re-invented thing in production, though. :)
What other good beginner advice is bad intermediate advice?
Okay, I'm stopping there because it's 5:30 am and I'm starting to babble. My mind is mush and open to the wisdom and experience everyone else has to offer on what you need to do to master Perl. :)
brian d foy <firstname.lastname@example.org>