5.17.6: "The seed used by Perl's hash function is now random. This means that the order which keys/values will be returned from functions like keys(), values(), and each() will differ from run to run."
Wait, what?
5.8.1: "Mainly due to security reasons, the "random ordering" of hashes has been made even more random. Previously while the order of hash elements from keys(), values(), and each() was essentially random, it was still repeatable. Now, however, the order varies between different runs of Perl." - http://search.cpan.org/~jhi/perl-5.8.1/pod/perldelta.pod#Has... , Sept 25 2003
When did that get undone?
Edit: Admittedly, I just tried to verify that this actually happened in 5.8.1 and none of 5.8.8, 5.10.0, or 5.14.2 appear to be actually randomizing key order, at least according to the keys function.
Thank you. That explains why my quick test (the obvious one) did not reveal anything in the older versions, I did not trigger the defensive behavior in my casually-chosen keys.
I'd be genuinely interested in reasons why I should learn Perl? I've been hacking away at some Perl interfaces lately but finding it difficult. It seems like a big language (I'm reading Perl Programming, which is certainly a big book).
So rather than just hacking away, is there a good reason I should learn the language properly and push through "Perl Programming" or say learning Ruby, Python or golang really well. Will Perl remain relevant in 15 years time? Or at least more relevant than other current languages with similar applications?
I should say that I come from a background of C and C++ programming, which may effect my mindset somewhat.
Installing a new version of Perl doesn't break your old code! This is not true for Ruby (I'm not sure about Python as I no longer use it much), which is a huge argument in favour of using Perl over Ruby.
Having said that, it's certainly possible to use all of these languages, and once you are good at one, it should take little extra effort to learn the others. All of them have their strong points and you'll probably regularly see code in each of them, so just learn them all.
I think that learning a scripting language is critical, but which scripting language is not. I used Perl for a while, then picked up Python, and I just adore Python. But pick one of Perl, Python, Ruby (or even Bash), and you're in good shape.
So, my actual advice is take a look at Perl, Python and Ruby, see which one excites you more, and commit to learning it for a while.
I've actually written small projects in all of them. The one that excited me the most was Ruby (and outside of scripting, golang). Ruby seems clean and small, which I like. But I also question the long term prospects of various languages. There seems to have been an explosion of languages in the last few years and I wonder how many will survive long term. Then I'm possibly just spoiled by C and C++, which just seem to maintain a (relatively) mainstream status.
Ruby, Python and Perl are older than you probably realize. Ruby started in 1993, Python in 1991, and Perl in 1987. I would not worry about their long term prospects: they've been around for a while, and I think it's a safe assumption they will be around for a while still.
Avoid decision paralysis. If you like Ruby, go with it. As I said, I don't think which scripting language you pick is all that important, but I think it's crucial that you know one.
> I'd be genuinely interested in reasons why I should learn Perl?
Perl 5 is handy, putting oft-used tools within arms reach. There are a lot of useful and mature Perl 5 modules available (at CPAN). The language has some warts you'll hear folks complain about. Perl 5 is considered a power tool for GNU/Linux & unix admins. Perl 5 infrastructure is great. Community support is great. The docs are great. For small scripts that involve text manipulation (with regexes) and working with the OS and with other scripts, it's very tough to beat Perl 5. The Perl 5 goals appear to me to be: be handy (sometimes providing more than one way to do it), be useful (bending the rules sometimes is ok, but try not to wreck up the place), and try to do what I mean.
Python is easy to pick up, is pretty uniform, and has a nice repl. You'll often find your Python code working correctly the first time you type it. You should be able to learn the language itself in a very short amount of time. The Python goals appear to me to be: be easy to learn, be mostly uniform. For more, open a repl and type `import this`.
The Perl 5, Python, and Ruby communities are somewhat different from eachother.
> Will Perl remain relevant in 15 years time?
Can't answer that, but I don't think Perl 5 is going anywhere for a while. Too much stuff runs on Perl 5.
I assume you mean "Programming Perl", aka the Camel book. (has a camel on the cover) In my experience the Camel book is better as a reference and you should use "Learning Perl" to actually learn the language. It's structured better for that task I think.
i'll wade in. note that i'm bigoted against perl, i'll say so clearly up front.
perl is better than shell scripting at automation, this much is true. perl has a strong use case there. had, i'll argue.
perl's idioms and structure promotes wonky, hard to read and hard to maintain code. it's true. get even the most die hard perl fan a beer or two and they'll wind up admitting it, sheepishly and quietly. but yeah, perl too easily promotes a mess. and when you have to maintain code, borrow code, etc .. readable code matters.
my biggest complaints about perl usually fall into the bucket that perl fans claim are its strengths. brevity, wit, etc. those wind up confusing almost all perl developers and users, in my experience, because you have to shift gears mentally to figure out wtf a piece of code is doing, what the developer intended it to do, and thus diagnosis is a pain. too many perl users just laugh and say, "yeah, i can't figure it out either." funny until you realize they're laughing about wasting your time. my second biggest complaint falls into the reusability of perl by its design. not so modular without some surgery (minor or major, depending). the C (or ruby or python) code that people trained on perl develop is, almost always, horridly inefficient.
CPAN is nice, i wish my language of choice (python) had as extensive a library so easy to use. and then you get into the mess of CPAN, which is littered with half completed, poorly documented modules. you wind up having two or three modules that do the same thing but are dependencies elsewhere or actually implement some promised functionality, etc. (cheeseshop, pip, etc yes ... but so much stuff isn't deployable like that is my point...)
the perl documentation library was a big selling point about 10 or 12 years ago, not so much now. everyone else caught up is my point. python, ruby, etc are now mainstream.
the "perl is everywhere" mantra used to be compelling, too. but python and ruby are just as widely available on systems and ... far more suited to long term development and maintenance than perl is. again, "better" languages for long term use and development are now mainstream. they weren't before.
i tend to sum all this up for someone with "perl sucks" or "don't use perl, it promotes brain rot". and no, i wont revisit my thinking. perl had its day, but i think the sun has set and rightfully so.
> "don't use perl, it promotes brain rot"
> and no, i wont revisit my thinking.
"I don't think this is the tool for the job, THEREFORE IT ROTS YOUR BRAIN! If you try to talk to me about that, I will chant, "I'M NOT LISTENING" over and over with my hands on my ears."
like another commenter said it's a discussion i've had before and haven't heard compelling arguments in favor of perl, so no i'm not interested in revisiting it.
that said it IS context dependent. i wont fire you for writing perl one-offs to maintain some systems or process some files. but i would argue against you for wanting to ship perl for the reasons i stated, with my main thesis being that your decision would jeopardize the project's future (assuming it had some future). i've heard the arguments before, like "you can write clean and maintainable perl" (which has analogues with "high performance java" and "long-term maintainable PHP"). the counterpoint is "sure, one can. but people rarely do. other tools promote those features more readily. pick one of them."
None of the reasons in your post would lead me to believe that Perl will 'rot your brain.' I didn't consider the majority of your post to be a flame/troll, just your honest opinion. It was the "Perl eats babies, and you can't convince me otherwise, nyah!" at the end that sort of irked me.
first, that some languages promote decent coding practices, while some actively encourage bad ones, either through crappy idioms, bad docs and a lousy community, or a lack of immediate penalties.
second, that where you spend much of your time coding has a significant effect on how you think about problem solving, consequences, etc. the practice of programming builds behaviors, modes of thinking, and values.
finally, if it isn't clear, i consider perl's features to be substandard to long term coding. the values it places on hacked up gimmicks (under the guise of "more than one way to do things" and "see! it's a one liner!") is a large part of this.
therefore "brain rot" in this case isn't equivalent to years of meth use, but rather a failure to appreciate crap designs, a failure to perform sound engineering when you write code, and a failure to think about the future of your code. i've seen the products of people who write code professionally, with them having learned through years of coding perl, and it is not pretty. lots of them.
keep calling me closed minded on the topic, i wont disagree. but the end result is that when someone tells me that perl is their language of choice and i'm hiring for a person to write code, it's a deep hole they have to get out of right away or i won't hire them, nor will many of the people i work with.
> the values it places on hacked up gimmicks (under the
> guise of "more than one way to do things" and "see!
> it's a one liner!") is a large part of this.
I'm curious where there is official documentation of Perl as a language putting emphasis on "See! It's a one-liner!" Sure Perl has things like Perl Golf[1], but I could equally point you to The International Obfuscated C Code Contest[2]. That doesn't prove that C as a language encourages you to intentionally obfuscate your code.
> when someone tells me that perl is their language of choice
> [...] i won't hire them
Perl isn't my language of choice, but I'm not sure that I want to work for someone that has such a view of the world. You're attempting to use a small piece of data ("I like to program in Perl") to extrapolate a whole lot about a person.
I am wondering why it is as it is. I think it may be because of perls abundance of short cryptic symbols that people feel somehow "proud" to 1) know what "while(<>){print if($.==2 || $& && !$x++); $.=0 if (/^--+$/)}" does and 2) to make it even shorter. Because with )(!§&/$?(=)% with hell a lot of meanings you could write entire perl scripts based on special characters.
A good example is http://www.nntp.perl.org/group/perl.beginners/2009/01/msg106... that i just found googling for perl one liners. The OP want's to know what this line does, the first answer is "you can make it even shorter!". Somewhat explains what i try to say ;)
Surely there is a lot of clean perl code out there but from my experience at customers (i'm consultant) what happens in-house in some development departments looks very very very different. Then again i've seen a lot broken, half-tested and unmaintained CPAN modules as well.
If you catch me using a butcher knife to get the cork out of a wine bottle, feel free to call me a fool. That shouldn't reflect badly on the makers of knives or corkscrews or wine bottles.
i've seen the products of people who write code professionally, with them having learned through years of coding perl, and it is not pretty. lots of them.
I've read a lot of code from people who've never learned a whit about programming. The lesson I take from that experience is that some languages make it easy for people who aren't interested in programming to get things done without having to learn much about programming or design or maintenance or coding standards.
That said, warnings and strictures should have been the language's default behavior for programs, not one-liners.
I think the argument might be that it's easy (easier than many languages) to slip into writing bad Perl, which promotes brain rot, it also sounds like it's a discussion he's had many times before... and perhaps isn't too interested in having again.
> it's easy (easier than many languages) to slip into
> writing bad Perl
Bad programmers will write bad code no matter the language. I don't like this idea that the language can somehow force people to write better code, or that it should.
See it that way: A bad programmer that can use such a variety of difficult to read symbols in a language (which perl clearly has) will just use all of them and the result will be horrible to read and understand. Also, often times python really has one obivous way to do a specific thing. Last time i was googling for some perl code (i think it was for some array handling) i have found at least 10 equally good totally different way on perlmonks in one thread. That's not bad per se but it promotes that every one uses another line of code and the bad ones choose the lines that may have unintended side effects but work in 90% of the usecases.
Using Perl will not rot your brain any more than any other programming language - balance is key: Look at other languages and don't become a one-trick pony.
If you look at the popular scripting languages (Perl, Python, PHP, Ruby, Javascript), Perl is the only one that gets freaking variable declaration right, so I'd be pretty careful when casting stones.
Exactly my feelings. Usually when i bring this up on HN it's just a heavy downvote and i gave up with it..
I must really say that the most horrifying times in my current job over the last years have not been critical situations like server failures or the like but the countless hours i spent on debugging and "cleaning up" the perl scripts of other people. Since then i made clear that i don't touch those anymore, someone else can waste their time. Especially when you spent more on fixing some script then it would take you to rewrite it in python..
I agree on several of your points. CPAN is kind of meh right now, since Python/Ruby/node.js are quickly catching up (or have passed it, even). You can quite easily find libraries existing in other languages and not in CPAN.
For me it now boils down to: sunk cost (if you're already familiar with Perl, it makes sense to keep doing it in Perl) and some features that Perl currently really excels at (variable scoping, regex integration, Unicode).
Yea, I kind of agree with some of this. My coworkers are big fans of Perl Critic because it enforces a standard (and I assume one that promotes cleaner Perl). But it often seem to me that the popularity for Critic is somewhat of a language failing.
I guess the argument could be main that it allows you to write hacky little scripts which could can then apply Critic too and slowly refactor into large projects. I'm not entirely sure I buy it.
No, whereas pylint promotes a good style, same space between function definitions, etc. this is something that is heavily missing in perl. Python also doesn't have (or actually it's almost never used) global variables and goto and all that crap that people still use in perl code. On of the first things i do when having to work on others perl code is to run Critic and first fix the myriad of warnings that come up. With Pylint it's typically just some minor stuff. With Critic it's typically some major stuff that's plainly wrong.
If you play around on *nix machines you'll want to know perl, at least how to read it and find translations for the things you can't read. You'll find it all over the place . . .
http://imgs.xkcd.com/comics/lisp.jpg
Let me tell you why learning Perl is important. You will never realize why Perl is important unless you start using Perl to solve the problems it was designed to solve.
The whole thing really is Perl doesn't have any real competition. And it has always been the case. And even the competing languages more or less end up largely looking like Perl. The biggest examples are ruby and Perl 6 itself. Python is not Perl's competitor. It took me a lot of time to understand this. Python is really for bored Java programmers, who are in some desperate need of dynamic power. And its great in that area. The language is readable, quickly learn able but what beyond that?
Unfortunately the problem with Perl is something similar to that of Unix utilities, unless someone shows you how to do magic with them. You never truly understand why and what purpose they serve. You can read as many man pages as you want, and think about a million use cases. But unless you see the next bearded sysadmin down your cubicle do a computation task in a combination of 6 text processing utilities sewing them with pipes, what other wise your Java programmer would have take six months to write- You will never truly appreciate the beauty of it. That's the reason why awk/sed/cut/tr/Perl et al have never gone away and probably never will.
I have seen Java programmers even wonder what 'Text processing' is, and found it strange that a regular expression can be anything more than 10 characters. The problem with most of those programming domains is most of them consume data in xmls or database records, where data is really reduced to a form where you don't have to do much to take it from there. That is why when you try understand why you must learn Perl from the perspective Java, you never get a reason. If you are learning any language from that Perspective, then all look the same. Because what else is there more to a language than decision statement, loops and ways of modularization?
Change this situation and that is when you see why languages like Perl are still alive and used all the time. Back in the 80's a person called Larry Wall faced a situation where he had to solve a problem where he needed to deal with a lot of adhoc text data, in many adhoc formats all needed to made sense of then sent through a traditional program and then made a report out of it. When he tried to solve it, he figured out he was using many unix text processing utilities and C programming language in shell scripts. He figured out that this made case for a separate programming language. When he sat down and put down some design in place he saw three details emerge out of it. Which are:
1. Manuplexity: Ability to manipulate things quickly. Like Text.
2. Whipiptitude: Ability to whip things up quickly.
3. Express yourself: To use principles of natural languages as applied to programming languages.
Finally there is a guiding principle: which really is Laziness, impatience and hubris.
If you really want to learn Perl, you should be reading books like Higher Order Perl. You should see tough text processing problems. You should see how you can rapidly prototype, even without the help of Modern IDE's which is rarity for most high level languages today. You should see how CPAN helps you avoid re inventing the wheel(s). You should see why things Moose and DBIx::Class are not there in other languages. You should see why the flexibility, extensibility leads to packages like Devel::Declare. You should see the testing culture.
Perl was born and raised among hackers, I mean the true bearded hackers. Who have experience of building things without reaching to the Google homepage every 30 seconds. If you have time, I would advice you to take some time out and read Tom Christiansen's essays back from those days. Take some time out to read JAPH and Golf contests from those days. You will see how much time those folks spent in doing the real hacker work, how many small succinct and yet extremely power solutions exist to so many problems you are likely to face everyday.
When you truly look at the use cases, you realize that Perl doesn't really have much competition in its stride. Even the languages designed to replace it look like Perl. Think of it like Greenspun's rule for Perl.
Even if you don't learn Perl I would definitely advice you to learn Unix utilities in quite some detail. That way you will realize and see a very different part of the world.
Thanks this was really useful. I use bash scripts, awk etc. fairly frequently but I guess rarely consider Perl for the same applications I do awk, perhaps I should.
Your comment really helped me understand where Perl should sit in my toolkit of programming languages, thanks again!
Indeed, the "Unicode support" in question is tracking changes to the Unicode standard itself. Perl 5 has long been one of the top-flight Unicode implementations (and not just implementation, but a community-wide set of best practices, documentation for use, etc.).
This is very thorough boilerplate code for dealing with all corner cases when using utf-8 data with Perl.
Instead of dismissing or dissing Tom Christiansen excellent post I would highly recommend reading into his The Good, the Bad, & the (mostly) Ugly presentation from OSCON 2011 [1] where he compares Unicode handling across mainstream languages and then see how this code (and Perl) shapes up in comparison.
In the meantime pragmatic Perl programmers can cover most of that utf-8 boilerplate with just:
use 5.016;
use warnings;
use utf8::all;
Or if you're like me and use perl5i [2] then its just:
Perl 5 was released in October 1994, so it's impressive in its own right that a) there is boilerplate you can add at all to get good Unicode support and b) that you can extend the language to support it using just boilerplate.
As the other comment mentioned, improvements by "default" to Unicode support do get included into later Perl 5 releases, but you have to let the compiler/interpreter know that you're buying into that so that it can reduce the boilerplate for you.
I mean Java had Unicode support right out the door in 1995. Not that it's a wonderful text processing language, but you'd think 17 years is plenty of time to catch up...
Supporting Unicode circa 1995 was good, but supporting an outdated version of Unicode incompletely isn't great. Per Tom Christiansen, JDK7 looks like the minimum required Java version to do modern Unicode correctly.
Despite being a pythonista at heart and a rubyist at work, Perl is still my go-to language for when I notice bash scripts getting messy and growing out of scope and destined to enrich my personal long-term unix toolset.
Yes, Perl is still very relevant today, for me and for countless others[0]
1. It's everywhere. By that I mean my code will work everywhere, without much thinking about it being installed, versions, or packages, or whatnot.
2. Ruby is too 'core' and needs too much additionals gizmos to be efficient
3. Ruby is concise and conventional, while implicitness and terseness are idiomatic of Perl
4. I'm no Perl monk but things in my head translate to code structure more easily in Perl for this use case (unix tools, text/binary/file content stream manipulation)
5. Perl strict mode
6. Occasionally, Perl suid support
7. Bonus: performance under Windows, or rather lack thereof for Ruby
Mind you, Ruby is extremely apt and I enjoy working with it (especially - but not only - with Rails), but this is more a matter of using the right tool for each problem.
The Perl core team are getting very good at timeboxing their development and releasing on time.