Hacker News Reader

If you enjoy Modern Perl, check out Higher Order Perl. It's a guide to functional programming in Perl and it's full of hipster-programmery goodness!

One way to measure the utility of a language is to see how its users perceive it. Both in the times of technology hype and otherwise. If I look at the Java world, I see most programmers come there because they need a job. I would program in Perl even if I would not be paid for it, just for the fun of it. Its like painting, or playing the piano, or drinking fine wine. I would like to do them, even there is no direct benefit from them.

Some languages get a special distinction that its users just love to program in it. Languages and tools like Lisp, C, Perl, Vi, Emacs etc they just never go away. Because someone somewhere is always discovering some goodness in them and is using it. And that process is timeless. These languages have managed to last for decades sometimes their adoption drops, sometimes it peaks, sometimes there is a constant line of usage. But they don't go away.

If you read the mailing lists during early days of Perl on comp.lang.perl.* and other writings by people like Tom Christiansen. You will see some unique tone of passion and excitement. The golf competitions, the discussions on one liners, the growth of CPAN, the occasional gems posted on Perl Monks. Great books like Programming Perl, Higher Order Perl and Modern Perl Book and Perl's extreme practical approach and emphasis of 'Getting things done'.

I guess for people on Unix platforms who had no other respite but to use C and a mash of other Unix text processing utilities. Or for Java programmers who have to write 10 classes to open a file and do trivial operations.

Perl is almost like giving Ice water to somebody in hell.

Although Python is taking more of these roles over time, Perl is still the thing that holds much of the internet's infrastructure and tools together, as well as internal servers, so it's a good thing to know if you're interested in all that.

Hmmm, Python has some great frameworks and is a great language.

But unfortunately, its almost like going backwards. Perl was invented because C + Unix text processing utilities together were not proving to be good for many tasks like Text processing, fast prototyping etc. Now using another C based language which doesn't do that, takes us back to the old problem again.

IMO, Python doesn't solve those problems. Regex support is really bad in Python and file operations are not trivial. There is too much of Java feel to it and it just doesn't have the feel of bash/perl/awk's scripting powers. Scoping is just not right. There doesn't seem to be a sane ways of evolution without breaking backwards compatibility. No equivalent of CPAN. And in many ways white space based parsing is not good for large and nested code blocks. It still doesn't have multiline lambda's. And its too much do-it-one-true-way kind of language, that sort of is a blocker for quick hacking. Because I like to write what I think, and not the other way around. The language is too verbose, and unlike Perl and Emacs there are no exponential productivity gains over time.

But I agree though, that Python does have some other good things. The OO syntax is neat, there are some great frameworks like Twisted and Django. And there is low barrier to entry which makes new bie feel just at home. The tutorials are just too good. And good standard libraries ship with it.

What you're saying seems handwavey and imprecise. What -- specifically -- is wrong with regular expressions and file operations in python? How is white space parsing "not good" for large and nested code blocks? (And why do you have large nested code blocks in the first place? Break those up! Those are hard to read regardless!) And how is there a "java feel"? What does that even mean? And in what way is python verbose? In my experience almost everything about it is almost as terse as possible, from the lack of needing curly braces to the minimalist syntax to naming things like "str" instead of "string"

'Programming languages teach you not to want what they don't provide.' -- Paul Graham

Its a lovely quote, but I've used Perl, I know what it provides. I've also used lisp. I just think if you're going to make a claim you need to back it up with something solid.

I gave you the solid things, but you give 'You aint gonna need it' kind of answer to everything Python doesn't have.

Coming to asking people forcefully shorten their code blocks. It's none of language's business to dictate its style on me and steal my freedom. A language is supposed to behave the way I want. That's its very purpose.

No multiline lambdas, No CPAN equivalent, Seamless interaction with shell like $output = `command` is absent. No first class regular expressions. Lack of ability to provide the command line power Perl gives. No switch statement, No equivalent of Perl's smart match operators, Nothing like Moose, Nothing like Devel::Declare, Parsing power nowhere even near to Python. And I can go on and on..

I guess these are solid enough things(at least for me)

"Coming to asking people forcefully shorten their code blocks. It's none of language's business to dictate its style on me and steal my freedom. A language is supposed to behave the way I want. That's its very purpose."

    if True:
        x = 1
        [copy the above line, paste it 999,998 times]
        x = 1
    
    print x

Python runs that and doesn't say anything about by block length; it just behaves the way I want and prints x (after a short pause). What are you talking about?

I was talking of programming style.

"No equivalent of CPAN."

Not exactly the same, but the standard library is extensive:

http://docs.python.org/library/index.html

There is pypi:

http://pypi.python.org/pypi

whose packages can either be installed directly or with pip.

> Regex support is really bad in Python

More verbose than Perl? Yes.

Bad?

    for x in re.findall(r'\d+', '12 drummers, 11 pipers'):
        print x

How is this bad?

> file operations are not trivial.

File operations are trivial and well designed. You know, there is much more to file operations than in-place substitute.

> There is too much of Java feel to it

Bullshit.

> and it just doesn't have the feel of bash/perl/awk's scripting powers.

Want it to feel like Perl? Don't bother - use perl.

> Scoping is just not right.

Scoping doesn't require your testimony.

    # foo doesn't exist
    if blah:
        foo = 'bar'
    # foo  exists if blah

This is perfectly fine.

> There doesn't seem to be a sane ways of evolution without breaking backwards compatibility.

When you change the semantics of the language, bozo syntax tricks don't help. Python 3 has semantic changes - syntax tricks aren't going to handle them.

> No equivalent of CPAN.

A large stdlib and PyPi.

> And in many ways white space based parsing is not good for large and nested code blocks.

Spaghetti will look like spaghetti, irrespective of whether block separators are braces or indentation.

> It still doesn't have multiline lambda's.

"still doesn't" sounds like it is something which is planned but not achieved yet. It is not something that Python designers are planning to incorporate in the language. Perl has multiline lambdas but I only used single line lambdas for sort, grep et al - anything larger than that becomes a named function. Unless you have blocks, multiline lambdas aren't used that much.

> And its too much do-it-one-true-way kind of language, that sort of is a blocker for quick hacking.

There can't be an objective explanation to a subjective statement.

> Because I like to write what I think, and not the other way around.

What's the other way round? I like to think what I write? So in Python you think what you write? That makes zero sense.

> The language is too verbose, and unlike Perl and Emacs there are no exponential productivity gains over time.

Complete, utter bullshit. There isn't a single language in the world where the productivity gains are exponential over a long time - there might be some when you are a beginner which gets capped pretty fast.

I am going to neglect the responses 'Bullshit' basically because that means you don't have an answer at the moment.

>>There isn't a single language in the world where the productivity gains are exponential over a long time - there might be some when you are a beginner which gets capped pretty fast.

Sorry to ruin your party. But there are languages where there are exponential gains in productivity with time. Languages like Lisp, Perk, Haskell etc. But I can't convince that to people who can't look beyond 'readability matters' and consider that as the Pinnacle of programming language design.

Coming to regular expressions they are first class citizens in Perl. That is basically what takes them to the next level. Now scripting on Unix has always been so much about Text processing. That is how it has always been. tr, sed, awk and dozen other text processing utilities. And then there is Perl. If you are not getting 'Text processing' thing correct, then you already loose the scripting race with a very huge margin.

Perl added a lot of new semantics to the language too given/when statements, smart matching operators etc. Python broke it for as simple a thing as a print statement. That is the whole point(The language is too brittle to evolve). And I'm sure any new thing in Python is going to go the same way.

And stdlib is no where near CPAN. Although I agree PyPi is doing a good job.

You are right that I must use Perl,And you are also right that Python is an awesome language. There are great frameworks Twisted, Django, NumPy. I've used Twisted extensively. In many Python ways is a great language for web development, for studying algorithms(because its better psuedocode), and due to its frameworks like Twisted - its good for network programming.

But I am sorry to say Python doesn't do well in scripting areas. And it so happens once you start there and go a little deeper you can just go on and on with Perl.

> I am going to neglect the responses 'Bullshit' basically because that means you don't have an answer at the moment.

Saying 'python feels like java' or 'exponential gains from switching language' is bullshit, and augmenting it with "calling it bullshit means you don't have an answer" is more bullshit.

> But I can't convince that to people who can't look beyond 'readability matters' and consider that as the Pinnacle of programming language design.

I don't remember mentioning 'readability matters' as being pinnacle of programming. You are responding to arguments you are imagining I am making.

> Sorry to ruin your party. But there are languages where there are exponential gains in productivity with time. Languages like Lisp, Perk, Haskell etc.

Get off your high horse. Lisp, Perl, Haskell aren't some exclusive club only you are allowed into. I have used Racket, Clojure, Python, Ruby etc. and if your gains are exponential accumulated over a period of 1 month between say Python and Racket, be assured you are a very lousy Python programmer. Let alone exponential, a gain of even 2x is quite a stretch. Why don't you put your money where your mouth is and show me your 'exponential gains'?

Sorry! I shouldn't have hurt your religion at the first place.

> Sorry! I shouldn't have hurt your religion at the first place.

In place of spewing this nonsensical shit, you could have quoted me an example of "exponential gains".

Out of curiosity, have you actually programmed in Lisp or Haskell? I am not saying you have, I am not saying you haven't; it's just that people who claim "fuck yeah macros - now I am 100x more productive" are mostly people who haven't actually used them and are regurgitating blogosphere bullshit.

Well I should have learned long back to not argue with impolite people.

This is one lesson I'm not going to forget.

> Well I should have learned long back to not argue with impolite people.

Sheesh. In this thread, there is nothing but platitude in your posts, and you are constantly trying to put down people(everyone else who doesn't agree with you about perl is a troll, saying that what you know about python is not correct is me defending my religion etc etc), and then you whine about people being impolite. Pot, meet kettle.

> This is one lesson I'm not going to forget.

Another lesson would be to actually learn and use Lisp before claiming exponential gains.

It's not impolite to say you are regurgitating blogosphere bullshit when you claim with utmost confidence "I am sorry to ruin your party but there are languages like lisp and haskell which results in exponential gains" when actually you haven't used either one of them.

What you consider as verbosity in Python might the 'being explicit' philosophy of Python. For example though the succinctness of crafting a Perl regular expression is alluring , I have found writing regular expression the Python way and referring the results in an explicit manner helpful in keeping the code readable.

The gain you mention (though the exponential part is debatable) is not limited to Perl alone[1]. Scripting languages provide a substantial gain in developer productivity freeing them from things like typing, faster debug cycle etc. and that gain is not limited to using Perl alone.

Check Table 1. in the reference I have mentioned, which compares the time it took to develop the same application in Tcl with time it took for development using in C/C++/Java. I think similar gain can be seen when one uses other scripting languages (say Python or Perl).

1. www.stanford.edu/~ouster/cgi-bin/papers/scripting.pdf

There is no hoopla outside of the echo chamber. If you want a truly amazing language I would suggest things like Haskell, Racket, Clojure, Erlang, Smalltalk, and/or Common Lisp.

If you're looking for more batteries included then I'd suggest Python or perhaps Ruby. Things are always going on in those communities and they've made some great killer apps like Blender (python), Rails (Ruby), Django (Python) and countless more. These languages are just as flexible as Perl, though they have their strengths and weaknesses in different places than Perl does.