Perl is dead

[FunkyMonk]

Long Live Perl!

I wrote a script today that tracks users of PerlMonks - I want to know how many active users there are. We've had ~170 non-anonymous monks visit in the 7 or 8 hours I've been running the script.

For a lazy Sunday, I think that's pretty good

Perhaps Perl isn't as dead as some would have us believe

Update:

 27 hours after the script started and 511 different registered monks have visited
2 days after I started logging and 730 different monks have visited

[grinder]

(sigh)

One Sunday afternoon does not a sample make. You can downvote Cop all you like (and I wish people would stop doing that, it's pointless, and merely pollutes $NORM), but (s)he does have a point. Perl is in decline, at least if you care to analyse some more meaningful statistics from the Perlmonks website itself.

Here are some statistics upon which to meditate: the number of top-level nodes created each year (where a top-level node is a SoPW, CUFP, Meditation etc.) I've drawn a bar graph as well, to make the trend more obvious.

year	Top level nodes	x500 posts
2001	12436	*************************
2002	12303	*************************
2003	11893	***********************
2004	11454	***********************
2005	11883	***********************
2006	8986	******************
2007	7950 (est).	****************

So the site is clearly in decline. To state the contrary is to confuse desire with reality. The question to ask, though, is how well can this be mapped to the overall health of the Perl community? I think the answer is... poorly. There are small but vibrant discussions over at use.perl, there's loads of traffic on all the Perl mailing lists I read, and there's a considerable volume of new CPAN releases each and every day.

That said, I've been coming here for seven years and it's still my favourite site on the web. I've become a much better Perl programmer, and a programmer in general, as a result.

[ctilmes]

So the site is clearly in decline.

Clearly our efforts to educate new users into using search to find previous answers rather than reposting the same questions are paying off!

[Nkuvu]

I've been using Perlmonks fairly often over the past few weeks -- more so than I have in prior months. However I have been using the search, and every single question I've had has been addressed already. Obviously I can't speak for everyone, but your observation is spot on for me.

[dpuu]

If we've answered all the questions that relate to the base language then that means that the base language is indeed dead. Just as the science of physics was "dead" at the end of of nineteenth century. Then something new comes along -- be it relativity, quantum physics, Perl 5.10, 5.12, or 6.0 -- and the new features stimulate new ideas, new idioms, and new discussions. If the resurrection has been successful the these ideas act as the catalyst for still more ideas (i.e. we shouldn't need a subsequent kick-starts).

[andyford]

My guess to explain the drop would have been that there's so many questions already answered here that there's a reduced need to create more nodes! Yes, plenty of questions are asked over and over, but even I use the search functionality occasionally.

To get more top level nodes, just invite the maintainers of some modules to move their support forums from mailing lists to perlmonks.

• Specialized questions about specific modules are the ones that often go unanswered here anyway.
• It's a waste of energy to sign up for a mailing list just to ask one question about a module.
• Many module mailing lists are so lightly trafficked that it's sometimes hard to get an answer to a question. At perlmonks, you can at least get a educated guess from someone that might spark a new direction of thought.

[blazar]

To get more top level nodes, just invite the maintainers of some modules to move their support forums from mailing lists to perlmonks.

I personally believe that this is not a bad idea in and of itself. But if implemented plainly like thus, it would be a real mess: PM currently is not an infrastructure apt at that kind of service. You know, for example, the Monastery's opinion on the creation of new sections. Said this, something that would work better in practice and I somehow long for, but with a desire most probably deemed to stay unsatisfied, is a series of alternative interfaces to the Monastery, beyond the web one.

Since you mentioned mailing lists, indeed the kind of alternative interfaces I was thinking of would be by mail and through nntp. In fact this has come out quite recently in clpmisc -in a thread (link @ GG) in which I manually "syndicated" a post from here- and quite expectedly there have been disagreeing opionons: Uri Guttman, for example wrote:

i am not there because of the web api (why no news/email gateway?)

implicitly implying that he would welcome such a gateway. But when I replied that "I would dream of such a beast," Tad McLellan commented:

Your dream is my nightmare. :-(

Several others agreed.

(Sad?) truth is that that newsgroup and this web forum are actually different media, with quite differing philosophies: they serve complementary purposes for people with different tastes. But as a general rule they're not in competition. And at least one clpmisc regular has recently become a monk too, thanks to that much controversial thread of mine!

So, at least in principle a dedicated newsgroup, mailing list and web-based forum even of the complexity of PerlMonks could be made to integrate nicely each other. But that would have been better planned in advance, rather than as an aferthought. Granted, there would be some obstacles requiring careful thought anyway. For example, the voting system: which may be implemented in the text-based gateway by means of a custom header or an especially formatted line in the body of the message -e.g. a '++' all on its own line-. More complex, even: of course on the web we have rich formatting, and some simple import routine may automatically put <p> tags around paragraphs from pure text and <code> tags around indented text which generally people uses for code. But there are potential problems, and more complex still: how to render the html formatted posts in the pure text environments? Well, many people use html-capable mail clients, but many don't and don't want to, and in usenet it's "no way" anyway. (Did somebody ever think of "nnhttp"?) One may adopt some translations that ought to work in most cases. In the long run such a circumstance may even have positive effects: it would put pressure on people using the web interface to use regular enough markup that it be translated easily to the pure text media, and on people using the latter, to conveniently break the text into paragraphs and indent the code by, say, two spaces.

End of brainstorming, for now...

[fenLisesi]

I think tye made some changes to site policy with respect to robots a few weeks after Are monks hibernating?. Would it be possible to compare the last three months, say, with the same period last year and previous years?

Update: Here is that information, number of top nodes for each month May-Sep, obtained from jcwren's PTAV (as referenced by grinder in Re: Are monks hibernating?), followed by the same data normalized to the month in 2002:

	'07	'06	'05	'04	'03	'02
Sep	665	650	958	921	966	951
Aug	802	733	995	1000	1022	1149
Jul	722	725	1024	1010	1176	1203
Jun	651	758	1046	1038	1095	1037
May	673	735	1088	1020	972	1142
SUM	3513	3601	5111	4989	5231	5482

	'07	'06	'05	'04	'03	'02
Sep	69	68	100	96	101	100
Aug	69	63	86	87	88	100
Jul	60	60	85	83	97	100
Jun	62	73	100	100	105	100
May	58	64	95	89	85	100
SUM	64	65	93	91	95	100

I am sure to have made some mistakes, but the general idea should stay in the same ballpark:

• There was a steep drop (by one third) going from '05 to '06
• No change this year compared to last year, though the meaning of last year's numbers depends on the state of the robots file at that time
Perlmonks top nodes frequency is only one indicator of interest in Perl. Other indicators include job postings and tiobe. Cheers.

[Gavin]

For a more accurate figure traffic / postings would need to be measured over the more popular Perl sites.
The figures quoted only show a decline in postings here.
That may not be the case elsewhere.

Long live Perl.

[stefp]

I would not fault Perl when taking declining perlmonks stats. Even if it is a social site, perlmonks is _so_ web 1.0. I hope some bright guy will use Perl6 and the Conway Perldoc spec to design a new shiny perlmonks.

[blazar]

I personally believe that PerlMonks in 1.0 times was more 2.0ish than quite a lot of other sites. Actually it was and still is different from most other fora out there - in a positive manner IMHO. Now that we're in 2.0 times I would say it is 1.5ish, but speaking as a person who's generally fond of 2.0's typical doodads, bells and whistles, I don't see any of them as being particularly fitting for PM nor do I see the latter to benefit in any particular way from any of them. Put it in another way: were PM "2.0", it wouldn't pretty much be PM anymore.

Granted, it's not such a black/white situation: occasinally I wish PM supported asynchronous votes and had collapsable nodes and nodelettes, perhaps with dinamically loaded info. But little more. OTOH there's been at least one attempt to bring such goodies to PM without touching its basic structure.

[KurtSchwind]

I'm heading back to Perl because I was forced to use Java for a bit and I can't believe that people are using that language for flow control and ETL in the unix environment. I ran a simple benchmark and Perl ran 30x faster than comparable Java code. Does this mean that Perl is 30x faster than Java? No. But in my situation, there is a clear advantage to switching back to Perl.

It's hard to beat Perl for small unix applications. (For possibly large values of 'small')

Update: Fixed some of the more creative capitalizations.

[thezip]

...oO(Perhaps you mean not so capitalized versions of *Perl* ?)Oo...

[Dominus]

I usually imagine that Perl will end up in pretty much the same place as COBOL. It will be dead, but it will still be wandering around, eating brains. The aging Perl programmer community will point at the festering, animated corpse and say "Perl is more robust than ever! Look at those maggots wiggling in the eye sockets!" And the rest of the world will make jokes like "No, Perl's not dead; it just smells that way."

I'd like to be wrong.

[webfiend]

I'd like you to be wrong, too. Any ideas for helping Perl avoid a fate that requires Perl developers to be dispatched by removing the head or destroying the brain?

[sh1tn]

IMHO you are not wrong. Never-the-less Perl will stay as hard link between procedural and oop.
And of course the one-liners, more or less synonyms for UNIX administration.

[apl]

I was talking last night to two friends who are both programmers. I asked what they were working on (C++, VB), and I said I was doing a lot of Perl. One mentioned that he also did apps in Perl. He never remembered the language, but after 30 minutes he'd again be coding like a demon. Then two months would pass, and the cycle would repeat.

The other simply smiled and said "Perl is the glue that holds the Web together". He does a lot of programming in Perl, but doesn't think of himself as a Perl programmer any more than he would mention in an interview that he speaks English.

[Aim9b]

So, perhaps a related question could supply either some validity to the premis, or at least some additional considerations as to the conclusion. "Why, when, & for What- do you prefer perl over language X?".

Personaly, I enjoy the "downward" scalability of perl as much or more than any other feature. When is the last time you needed a simple conversion program and didn't want to bother with the whole .NET framework or the MS Foundation Class, etc. What's the last BATCH file you scrapped because it fell just short of the required capabilities? Yes, I know these are more Win issues than *NIX, but maybe perl is not dead, but morphing... into an OS independent utility language, rather than a full-scale major project developement language.

Give me a user suported fast, functional tool ANYDAY, over a GATES-ified behemoth that is only half-fast at best. These are just the things that attracted me to perl in the first place, lets exploit them. Does the world really need an OO capable scripting language? I'd suggest that better marketing may prove more benificial that additional features. Juat a thought..... Thanks.

[dsheroh]

Erm... Morphing? Hasn't Perl been "an OS independent utility language" pretty much from the start?

[Aim9b]

OK Perhaps morphing was a poor choice. I don't know how long it's been available on Windows, but there are still a few places that it doesn't run. So I really meant more of "Broadening it's scope" type of change.

Also, It looks like perl is gaining features, faster than it's gaining additional platforms. Not that I thnk this is right or wrong, but there's just something about an "object oriented 'glue' language" that sounds oxymoron-ish. Make no mistake, I love all the features, but if we're debating whether perl is dead (or dying), perhaps the "cure" is to get it running more places.

One of my dream projects when (if) I retire, and the main reason I hang out here at the Monastery, is to someday have enough smarts to port perl to a new platform. There's a chance it will get there ahead of me, but the training I gain in the meantime is invaluable.

So, thats what I meant, sorry about the confusion....and the rambling. As of todate, my prose is more proficient than my perl. ;-) Aim9b.

[zer]

Its amazing how we can relate the life of a language through a website! How many other internet communities are there out there that revolve around a language? Just the fact that we can take usage stats off of this site to demonstrate how people use Perl in this day and age is incredible.

Would this mean that Perl will not become officially dead until this website reaches 0 users per day? I don't know. What would the requisites be for a dead language? Lisp has fooled us all with its death but is still lingering. Cobol and assembly have understandably come to an end. But even so not really. There is a large demand for legacy coders today. Even pre-standard C coders are needed.

I suppose it is fair to say that even dead languages are never really dead. Perhaps just reborn?

So therefor, Perl will never die. Just become reborn.

[syphilis]

2 days after I started logging and 730 different monks have visited

These were probably the 730 registered "Anachronistic Monks" who misread the subject as "Paul is Dead".

Cheers,
Rob

[Anonymous Monk]

Just thought I should mention that, although I'm a registered user, much of the time I simply browse PM, and don't bother to login. Oftentimes, it's in an inconvenient place and time for that. . .like right now. So I think measuring registered-monks traffic doesn't neccesarily indicate the value of the site in general. Just my $.02.