Perl 5.16 and Relevance [LWN.net]

Perl 5.16 and Relevance

Posted Sep 14, 2011 9:11 UTC (Wed) by kragilkragil2 (guest, #76172)
Parent article: Vincent: Perl 5.16 and Beyond

Just out of curiousity: What was last big FOSS project that was started and uses Perl?
I hear about stuff written in C(++), Java, Javascript, Python and PHP all the time. Sometimes I hear about Rails projects, but Perl .. I nearly never read or hear about new well-known projects written in Perl (besides the occasional small Wiki engine).

Is Perl just for FOSS legacy applications?

to post comments

Perl 5.16 and Relevance

Posted Sep 14, 2011 11:37 UTC (Wed) by FranTaylor (guest, #80190) [Link]

Perl is for all manner of internal plumbing for your site.

Use it to generate reports, to auto-configure your application when you reinstall it. Use it to fill in the missing bits of functionality when you smash together a bunch of existing programs.

Perl is duct tape for the system administrator. You wouldn't want to code up a word processor or a database in it, but it sure does come in handy. Even if your app is written in something else, Perl can be the fastest and most expedient way to add that missing functionality that pulls the whole room together.

Perl 5.16 and Relevance

Posted Sep 14, 2011 13:29 UTC (Wed) by renox (guest, #23785) [Link]

I don't think it counts as a big FOSS project, but many tools in Mandriva were coded in Perl.
Not sure that it is a glowing reference though..

Perl 5.16 and Relevance

Posted Sep 14, 2011 14:14 UTC (Wed) by dskoll (subscriber, #1630) [Link] (8 responses)

As others have mentioned, Perl is used with in lots of plumbing. Many Debian system admin scripts are written in Perl (though the quality of said scripts is uneven at best.)

High-profile Perl projects include SpamAssassin, Catalyst (on which a bunch of web sites run), RT (a ticket-tracking system) and Slashcode (that powers slashdot). Admittedly, these are not new projects, but they are actively maintained.

Perl is also used a lot in bioinformatics, I believe. Of most interest to me, my company's (non-FOSS) products are written mostly in Perl.

Perl 5.16 and Relevance

Posted Sep 14, 2011 17:42 UTC (Wed) by k8to (guest, #15413) [Link] (6 responses)

Debian has been moving deliberately towards python from perl for their administrative tools. They don't foolishly abandon stuff just for language purity but the drift is there.

I'm intrigued that bioinformatics is still solidly on the perl train. It always seemed a surprising choice to me. Perhaps it will end up like fortran for physics -- used forever in that space.

Perl 5.16 and Relevance

Posted Sep 14, 2011 17:57 UTC (Wed) by andrel (guest, #5166) [Link]

I know bioinformaticians who still use Perl. But there's a strong trend towards using Python or R.

Perl 5.16 and Relevance

Posted Sep 14, 2011 21:57 UTC (Wed) by dskoll (subscriber, #1630) [Link] (1 responses)

Totally useless and meaningless statistics from my Debian Squeeze system:

$ echo /usr/bin/* /bin/* /usr/sbin/* /sbin/* | wc -w          
4619

$ file /usr/bin/* /bin/* /usr/sbin/* /sbin/* | grep -c shell.*script      
621

$ file /usr/bin/* /bin/* /usr/sbin/* /sbin/* | grep -c perl.*script
420

$ file /usr/bin/* /bin/* /usr/sbin/* /sbin/* | grep -c python.*script
198

So shell (typically Bourne shell) is ubiquitous, Perl looks solidly entrenched and Python is indeed up and coming.

Perl 5.16 and Relevance

Posted Sep 15, 2011 3:46 UTC (Thu) by k8to (guest, #15413) [Link]

I was going on some wiki and email policy statements as well as seeing a few key tools move to c or python. python for example is the recommended interface these days to apt.

But there are plenty of debian developers who do not view a sort of vision statement like that as very important so yes some perl tools will remain forever as sprout anew.

Perl 5.16 and Relevance

Posted Sep 15, 2011 22:58 UTC (Thu) by BenHutchings (subscriber, #37955) [Link]

Debian has been moving deliberately towards python from perl for their administrative tools.

There has been no such policy decision. Further, a subset of Perl (the perl-base package) is 'essential' (guaranteed to be installed) while no such subset of Python has that status. (Which I regret, as I would much prefer to use Python for complex maintainer scripts.)

Some utilities accompanying dpkg were written in Perl but have been rewritten in C so that 'Emdebian' installations can leave out perl-base as well.

Perl and bioinformatics

Posted Sep 16, 2011 13:55 UTC (Fri) by Kluge (subscriber, #2881) [Link] (1 responses)

I think part of the reason Perl is used so widely in bioinformatics is historical. But I think it's also because much of bioinformatics is text processing, and what is better suited for that than Perl?

Perl and bioinformatics

Posted Sep 16, 2011 15:34 UTC (Fri) by rgmoore (✭ supporter ✭, #75) [Link]

I think this is essentially correct. The problem in Bioinformatics, as in many areas, is that much of the early work was done in widely separated groups who weren't coordinating much. The result was a bunch of really good tools that don't play nicely together because they speak different languages. Rather than try to rewrite everything to work together perfectly, they stuck it together with a glue layer written in Perl. And of course once you have a setup in Perl that's the only thing that understands all the different languages, it's the natural choice to build new tools.

The result is that Bioperl now has literally hundreds of modules. Many of them are rather special purpose, but taken together they give you a very nice toolbox for solving your Bioinformatics problems. They may be a bit heavyweight- for some simple problems I find it's easier to write my own, lightweight solution than to read through the reams of Bioperl documentation to find the tool that does what I want- but that's the price you pay for the power and flexibility.

Perl 5.16 and Relevance

Posted Sep 15, 2011 3:07 UTC (Thu) by sumanah (guest, #59891) [Link]

Just adding another actively maintained Perl project to your list: Dreamwidth. http://wiki.dwscoalition.org/notes/Dev_Getting_Started for more info.

Perl 5.16 and Relevance

Posted Sep 15, 2011 6:20 UTC (Thu) by szabgab (guest, #80245) [Link] (1 responses)

AFAIK Most of the open source Perl stuff consists of libraries allowing one to build large in-house applications. Both web applications and automation stuff. See CPAN for details.

For applications you could take a look at Perl Dancer web framework and at the dancefloor where they show a number of projects written using Dancer.

Another option is the Mojolicious web framework though I am not sure where do they list the projects written in it.

Finally let me point to Padre, the Perl IDE. While being an IDE for Perl is a bit self serving, it is a large open source projects written in Perl recently. (disclosure: I started this project)

Perl 5.16 and Relevance

Posted Sep 21, 2011 21:53 UTC (Wed) by sebastian (guest, #4910) [Link]

Perl's ease and speed of use are important. A lot of device vendors will ship sample developer kits with Java and C, but these take quite a bit of time to decipher. If I was developing a fancy expensive interface I would take the time to dig thru the developer kits. But a quick search on CPAN may reveal modules that will get you up and running in a couple of hours. Try that in Java. You could waste a week coding up a working example and have less functionality than in Perl.

FOSS Project in Perl

Posted Sep 17, 2011 22:04 UTC (Sat) by plaxx (guest, #53703) [Link]

Another FOSS Perl project: PacketFence (network access control). We announce our releases here on LWN (http://www.google.com/search?q=site%3Alwn.net+packetfence).

The choice of Perl was done before my time so I can't tell why it was decided that way but I can say that it has proven to be very productive (developer time wise), easy for users to do customization and easy for us to implement quick on-site fixes or apply patches to clients (because no recompilation required). Also, network administrators seems to be speaking Perl more than anything else so they are familiar with it.

I am not saying that Python doesn't apply to all the above (I believe it does) but so does Perl and a project with that history won't be reimplemented just because the cool kid around the block is python these days. It would not be economic to do so.

Full disclosure: I am PacketFence's lead developer living the dream of doing open source for a living at Inverse (http://inverse.ca).