Hacker News Reader

Comment by microtonal | Hacker News Reader

And another metric shows that Perl continues its years-long decline:

http://www.tiobe.com/index.php/content/paperinfo/tpci/index....

This is one of the factors why we have replaced Perl in our curriculum with Python. The other being that Python is far more fit for education.

View on HN

For stuff like string handling, text processing and even variable interpolation, Perl blows Python and Ruby out of the water (and despite criticizing Perl's regex, Python ended up copying it almost identically). Much of this power stems from Perl's much-derided use of sigils (e.g. $scalar, @array, %hash) and context.

Personally, I find Perl's $header2 = "Dear $title $name,";

to be more readable than Python's header2 = "Dear %(title)s %(name)s," % vars()

Another benefit to using sigils is that it's immediately clear when reading code what type of data is being referred to, while in other languages you often have to read back through the code.

On the other hand, Python has done a great job in positioning itself as the new BASIC (in a good way), and so of course Python gets much more exposure in the educational setting and thus has greater mind-share amongst younger coders.

Python is simpler and easier to learn well than Perl, but nevertheless Perl is a very powerful and useful language that takes a lot of undeserved criticism (e.g. see the comment of wyuenho in this thread).

Incidentally, I found the Google Trends link hilarious, as all the news snippets on the right of the page refer to people called Ruby and actual Python snakes (hey, maybe people are searching for things other than programming languages..)!

I think what you have said is quite fair. However, I must clarify that I wasn't criticizing Perl's use of sigils, I was criticizing Perl's use of punctuations. There's big difference. Sigils does indeed make the types of variables clear, tho I find that context alone is enough. Objects in natural languages don't have to have types in their spellings to convey their categories after all. The use of genders in some languages is superfluous, Chinese doesn't require that.

All I'm saying is, Perl requires too much more mental power to process than Python because of it's large vocabulary and inconsistent grammar rules. It is really a natural language, but a natural language is not needed for programming.

Yeah, punctuation, wow. I mean, here's a sierpinski triangle generator:

32.times{|i|i.upto(31){|k|putc~k&i>0?32:35};puts}

Oh, wait. That's Ruby. I meant to say:

<?for($n=33;$k=--$n;print~\xf5)for(;$k;)echo-$n&--$k?~\xdf:~\xdc;

Oh, sorry! That's PHP. My bad... here we go:

print map($%&$_?$":'#',0..~($%=$_)),$/for-32..1

There we go... You're right, Perl is so much worse!

The point is, of course, you can write ugly code in any language, or you can write beautiful code in any language. Python actually is fairly legible nearly all the time...

n=32

while n:print''.join('# '[-n&k>0]for k in range(n));n-=1

but that comes at the cost of added verbosity and reduced flexibility . Whether that's better is a matter of taste, not of fact.

I'm sorry, but which grammar rules in Perl 5 are inconsistent? I've taught Perl 5 professionally, and while I certainly do acknowledge that Perl isn't optimal for learning, "inconsistent grammar" is not on my list. It does have a more-complex grammar than, say, lisp, but it's still an extremely consistent grammar, as far as I've been able to tell.

I also disagree with the claim that a large vocabulary in Perl is a notable issue. The size of the Perl grammar and language is sufficiently smaller, in my experience, than the size of all of the library APIs that I have to deal with when writing Perl programs.

For stuff like string handling, text processing and even variable interpolation, Perl blows Python and Ruby out of the water...

I agree on Python. I disagree for Ruby. It is a tossup. In most cases they are equivalent, Perl has autovivification, it is easier to interpolate functions into strings in Ruby. All in all it doesn't matter.

If you use the TIOBE index as a measurement for language popularity you shouldn't be controlling a curriculum related to information technology because it indicates you don't have high-school level knowledge of statistics. TIOBE is based on "searches for a given language", there are all sorts of factors that can skew the outcome pretty badly. See http://www.tiobe.com/index.php/content/paperinfo/tpci/tpci_d....

You must have misread my comment. Popularity (as in what typical future employers require) is one of the factors (besides fitness for education and program consistency), TIOBE is not.

The TIOBE reference was to show that there are nearly always metrics that show your point.

But I guess it is easy to stomp on people based on presumption. Let's just stick to arguments, rather than poisoning the well.

If you are using TIOBE to make decisions on what to teach, I feel bad for your students :)

That's true, but having said that, it's well known (or at least often joked about) that Perl makes a poor first programming language but a fine _last_ programming language.

Jokes aside, I think the sentiment has some strong truth - if for no other reason that there are some very heavy assumptions in the standard documentation that the user is well versed in Unix, shell and C libraries. "perldoc -f printf" and "perldoc -f sprintf" are _not_ what I'd expect first-programming-language learners to have to work with.

Unfortunately, the usual refrains of "patches welcome" are unlikely to solve this problem, because as a 2nd or 3rd (or 17th) language, it's precisely the documentation I want.

That's a caricature of my comment (given the smiley this is probably intentional). The primary motivation was educational fitness, the second consistency within the education program, the third popularity. Of course, not popularity according to the TIOBE index, but to what typical employers require. Of course, in the market C# and Java are more popular, but popularity is just one factor.

Reply on news.ycombinator.com