Tired of "Perl is dead" FUD ?

brian_d_foy on 2007-09-14T00:46:00

Yeah, me too. But I'm a capitalist, so I like to see what people are writing checks for. As I'm fed up with the various unsubstantiated claims, I've whipped up a little graphic of job posts on Dice that will hopefully cheer you up. I'll try to keep it updated regularly. If nothing else, it'll give us all something to watch as we're overtaken by our Ruby and Python overlords.


Why is Perl's bar so tall?

brian_d_foy on 2007-09-14T00:30:34

I would tend to think that's right, but it's counter to what a lot of other people report. Is it just the way different people count?

It would be really cool to see a Venn diagram sorta thing to see how they all overlap, but I guess that would be too many dimensions. :)

Any chance of sharing the code? Maybe as an article for The Perl Review? :)

Re:Why is Perl's bar so tall?

Alias on 2007-09-14T03:01:03

I'd say the reason Perl is to tall is because so many jobs have "this, that, the other.... and Perl".

That's part of the reason jobs.perl.org became so popular, because on jobs.perl.org you know the jobs are PRIMARILY about Perl, not just incidentally involving it.

Re:Why is Perl's bar so tall?

brian_d_foy on 2007-09-14T03:37:22

I think that's probably true, and maybe it's just been buried in all the jobs that listed Java in the same way.

That's the tricky thing about this sort of analysis: what are you going to actualyl be doing when you get the job? There might be a Perl keyword and then they switch it up on you so you're really doing PHP . :)

Re:Why is Perl's bar so tall?

cwilbur on 2007-09-14T14:45:45

When I was recently looking for a new job, about one in three positions I considered wanted Perl programmers because they had a massive legacy system written in Perl that they wanted to port to Java.

Before I put any faith in this graph, I'd want to know about how it's determined that Perl (or any other language) is primary or secondary, and what features of the dice.com job posting makes them especially representative of the actual industry. I'd also want to know how duplicates are screened -- how does the software distinguish between ads for two different jobs and two ads for the same job, one posted by the HR contact at the company and the other posted by a recruiter?

Re:Why is Perl's bar so tall?

renodino on 2007-09-14T15:33:52

Any chance of sharing the code?
I've posted it to perlmonks.

Its pretty simplistic, so I don't think its TPR-worthy.

Many will question the "sampling" technique, but I'm assuming the spurious datapoints are as likely in PHP/Python/Ruby samples as Perl (which is why I added the "Mixed" datapoint for each language). As importantly (albeit subjectively), I think job listings are a much better metric than the usual book sales or (perhaps worst) web search metrics.

And yes, I pondered a Venn diagram, but DBD::Chart can't do that yet ;^). I thought about using a quadtree, but that would likely have made an even more pronounced difference between Perl and the rest (ie, one huge square for Perl, and 3 small squares for the rest). I hope to keep a log of the daily results, and eventually create a trend line for it, but that'll take a much bigger sample (as the indeed.com 2 year results indicate).

As always, there's lies, damn lies, and statistics. But its nice to inject some hard numbers into religious debates.

Only Perl Jobs

naveenbala on 2008-02-12T18:12:20

brian, to answer a part of your requirement, we have a tool Market Statistics that you can compare skills based on location and industry type and it will give the monthly job demand for 2007 and it also compares median salary. http://www.odinjobs.com/StaffIT/MarketOverviewServlet
If you put (perl not java) you will get jobs that asked for perl and does not ask for java at all. you can compare (perl not java) versus perl and see what is the overlap of jobs between the two skills. Unfortunately, the query supports only 1 "not" and hence you cannot string together nots to eliminate other skills also. Hope it helps

And jobs.perl keeps growing...

Alias on 2007-09-14T04:49:35

http://jobs.perl.org/about/stats

Per city

jonswar on 2007-09-14T09:11:18

And just for fun, here's the # results per city for the top twenty cities:

  757 New York, NY
  211 San Francisco, CA
  191 Chicago, IL
  156 San Jose, CA
  151 Jersey City, NJ
  120 San Diego, CA
  93 Seattle, WA
  93 Mountain View, CA
  89 Atlanta, GA
  85 Santa Clara, CA
  82 Los Angeles, CA
  66 Philadelphia, PA
  61 Irvine, CA
  60 San Mateo, CA
  57 Dallas, TX
  52 Minneapolis, MN
  50 Palo Alto, CA
  33 Charlotte, NC
  30 Portland, OR
  27 Bellevue, WA

Not growing in my town

jbooe on 2007-09-14T12:19:55

I live in a fairly large metropolitan area in the Midwest (US). Here, as far as advertised jobs, PHP is king, as well as the usual Java/.NET.

Most jobs I see listing Perl have it in a generic list of requirements or for shops that need Perl programmers to help them migrate away from Perl. Once again, I'm only referring to advertised jobs.

As a job seeker/contractor, .Net, Java and PHP are the jobs that people are writing checks for. Perl is almost non-existent, and I've come across more Ruby jobs lately than Perl.

As an employer, we spent months trying to find a single Perl developer.

So sad to say, where I live Perl may not be be dead, but it's hard to find any signs of life.

Perl isn't dead, it's merely sleeping

djberg96 on 2007-09-14T19:12:22

Cobol returned 1287 hits, more than some of the other languages you showed there. Does that mean Cobol is "alive"? I wouldn't say so. Not in any "community enthusiasm" or "upcoming features" sort of way.

And, as others have pointed out, Perl is usually an auxiliary, not primary, job skill that employers are looking for. I think I see it listed for just about every sys admin position, for example.

Now, before anyone gets their Perl Panties in a bunch, I'm not suggesting that Perl is anywhere near where Cobol is today, but I think it's safe to say (in that anecdotal sort of way that chromatic hates) that the enthusiasm for it is in decline. Whether or not Perl 6 can reverse that trend remains to be seen.

Re:Perl isn't dead, it's merely sleeping

chromatic on 2007-09-14T20:59:45

I think it's safe to say (in that anecdotal sort of way that chromatic hates) that the enthusiasm for it is in decline.

That depends on your definition of "safe". If you mean "here is a stack of statistics and here is the research methods and raw data", then we agree on the word safe. If you mean "two 14 year olds emo kiddies with stupid pointy-in-the-middle haircuts and eyeshadow knocked on my door to try to sell me on DHHism, and no one's done that for Perl in a couple of years, but I leave the proof as an exercise for the reader" then I'm not going to equip a bunch of liberated but still angry chimpanzees from the monkey farm down the street with ball-peen hammers and point them in the direction of your kneecaps, but I might still ask for clarification.

In the latter case, you're mostly safe... but I will probably tease you on the Internet.

Re:Perl isn't dead, it's merely sleeping

renodino on 2007-09-14T22:04:03

Does that mean Cobol is "alive"? I wouldn't say so

Actually, I wouldn't call in Father Murphy to give COBOL its last rites just yet. And if COBOL's survivability concerns you a great deal, you may want to withdraw all your funds from the bank, as much of them are still managed by COBOL based systems. Not to mention how much of your government may be dependent on it.

I'm also a bit puzzled by those that seem to demand a level of data cleansing of the Perl numbers that isn't required of any of the other languages. Perhaps anecdotal random sampling is a better data collection technique ? "A guy I work with read an article about a guy whose sister dated a guy who said Perl is dead, and Ruby is the future." Unfortunately, it's difficult to quantify such data.

But it does raise another issue wrt to "vitality". I've vague notions of trying to collect metrics on the number of daily (or maybe weekly or monthly) CPAN uploads. Alas, I'm uncertain how to get comparative numbers for the other languages, since their equivalent archives seem a bit scattered or sparsely populated.

Perl isn't sleeping. Rather, it has matured. It has work to do, and doesn't really have time for the Little Jack Horner act anymore. After all, someone has to be the duct tape of the internet; otherwise, all those shiny new Rails and PHP apps with their sexy gradients and rounded corners would be all dressed up with no place to go!

Re:Perl isn't dead, it's merely sleeping

djberg96 on 2007-09-14T23:19:33

Oh, I'm well aware that COBOL will probably be around long after I'm retired. But then, I'm talking about community vitality. There's really nothing to look forward to with COBOL beyond a paycheck. They still teach COBOL in the US Military, btw.

Regarding the data cleansing, I think it's a fair question to ask for any language that's often used in an auxiliary role. If you want verifiable data I guess you would have to analyze the job descriptions. Personally, I think jobs.perl.org is a better (or at least easier) metric than dice.com.

Regarding CPAN, it's funny you should mention that. Tom Copeland recently blogged that RubyForge had over 20,000 registered users. By comparison I count only 6148 on CPAN (someone correct me if I'm wrong), so CPAN uploads may not be a great vitality metric if it's relatively few people doing most of the heavy lifting.

Re:Perl isn't dead, it's merely sleeping

chromatic on 2007-09-15T00:08:38

om Copeland recently blogged that RubyForge had over 20,000 registered users. By comparison I count only 6148 on CPAN...

To paraphrase Sam Tregar, that's kind of an Apples to Oranges comparison (even with your followup correction).

Re:Perl isn't dead, it's merely sleeping

djberg96 on 2007-09-15T02:06:31

Hah, fair enough. I'm not sure how you count the total number of Perl devs who contribute to Perl libraries but aren't listed. It's probably impossible without manually checking README and/or CHANGES files.

I've been doing some more number crunching for both CPAN and RubyForge, and it's been an interesting exercise. I'm going to save the results for an independent blog post. However, I'll leave you with this hypothesis:

The existence of a collaborative development environment for a given programming language results in more collaborative development among programmers of that language.

Re:Perl isn't dead, it's merely sleeping

chromatic on 2007-09-15T02:50:17

Does "collaborative development environment" include IRC, mailing lists, and Usenet or is it solely the purview of web fora?

I'm curious to see your research and conclusions, but the CPAN predates SourceForge. In my mind, that's an important distinction between collaborative Ruby and collaborative Perl development. (I might also suggest that Ruby's main driver skews a lot of new Ruby developers toward the web, where even Perl 5's various waves of popularity included a lot of system administrators who didn't do much with the web.)

Perl is dead?

awesomess3 on 2007-09-19T05:59:50

I think we need some more data....

$: for i in perl python java ruby php cobol ; do echo -n "$i -" ; apt-cache search $i | grep $i | wc -l ; done
perl -1560
python -720
java -426
ruby -375
php -245
cobol -1


I think this says that Perl is extremely mature. People are not as excited about it because it's old (that is, Perl 5), but that doesn't stand for anything.

The reason MacOS software developers are so successful is because everyone is developing for Microsoft Windows. There may be only 2 opening for Perl, and 5 for PHP that are for job openings. But think about, there are 293438 PHP developers out there and only so many Perl developers.

The less enthusiasm for Perl, the better it looks for perl developers finding jobs IMHO.

</optimism>