Perl Survey Report

All the raw data and analysis code for this report is available from github Summaries of findings for each category are given below. Please click on the link in the heading for graphics/data tables for each section of the questionnaire.


Of the 4800 respondents, they were 96% male, 3% female and 0.63% "other". Age distribution did not differ by gender - the peak age of Perl programmers is 30-39 (40% of respondents). The next largest group were programmers between the age of 25-29 (22% of respondents).

Countries of origin/birth

In this section we look at programmer country of origin and country of residence. This data is fairly complicated, and I encourage you to examine the data table for details. The top 5 countries of residence in order were USA, Germany, UK, Canada, Russian Federation and Australia.


Perl programmers are a well educated group. The majority have either an undergraduate degree or a postgraduate degree. By far the most common majors studied was computer science. Engineering, science and mathematics were the next most popular majors.

Perl related education

The vast majority of Perl programmers are self-taught - a few more but not many respondents were taught another open source programming language.

Employment info

Rather than describing income and employment sector information here, I encourage you to look at the graphical data on income, and the tabular data on industry sector. It's interesting to note that while Education was the least popular major topic studies, it's the 4th most common industry sector to work in.

Programing Experience

The questions here were:

Update: estimating the characteristics of less committed perl programmers

To get an idea of what those less committed to Perl look like, we selected the respondents who reported that Perl was their 3rd to 5th most commonly used programming language, and performed the same analysis with this subset. Graphs are available here. The results seem to indicate a much broader spread of time programming perl, which might suggest that a programmer who finds Perl early on in their programming career is more likely to stick with it than one who finds it later on.

Information about Perl

The majority of respondents reported using Perl version 5.8.x or 5.10.x. In terms of Perl usage, Perl 5.6 is used less frequently than Perl 5.12. 3% of respondents are still using Perl 5.005 and 2.5% are using version 5.4.

Operating System Info

This question summarises the number of people using a large number of different operating systems for development and deployment. While Ubuntu Linux is the most popular development OS, Redhat/CentOS is the most popular deployment OS. Mac OS X is common for development (20%) but relatively rare for deployment (8%).

Perl on Windows

ActiveState Perl remains the most popular environment on Windows.

Programming Language Information

This complex data summaries the popularity of a large number of programming languages. After Perl, the next most popular languages are C, Unix shell languages, Javascript, Java, Python, PHP and C++. It's worth looking at the data table for this, as it shows the frequency of use (1st-6th most commonly used) as well as the raw popularity of the language.

Editor information

37% of users use Vim as their primary editor, 15% use emacs. Vi/Vim variants are the next on the list followed by Notepad and TextMate. 2% of respondents use Padre as their primary editor. It's worth looking at the data tables, as it shows the primary editor, as well as secondary editors used.

Community involvement

Most respondents are not subscribers to Perl Mongers lists, and most do not subscribe to Perl technical mailing lists. Most respondents do subscribe to a range of non-perl technical lists. Most respondents, although they have a local Perl Mongers group do not attend, and most respondents never attend Perl conferences.

Perl Media

With the demise of, this list is now outdated, but at the time this was by far the most popular Perl news source. The new site has experienced a rapid rise in popularity with 800 respondents using it as a news resource (the third most popular source after Planet Perl).


Around 600 users have contributed 1 or more modules to CPAN, and mostly these are not handed over to others for maintainership

Bug Reporting

This section summarises the frequency with which bugs are reported and fixed in Perl and CPAN modules. Participation ranges between 5-25%. It's worth looking at the graphs to see respondents reported bug-reporting and bug-fixing behaviour.

Opinions about Perl

Here we see that while respondents attitudes towards Perl are generally positive, employers attitudes are seen as less positive.

We also asked three open-ended questions about Perl - the best thing about Perl, what frustrates you most about Perl, and In a few words, describe Perl. The Leximancer diagram below has interesting data about how people's responses to questions cluster together. To examine this data more closely have a look at the csv file with the raw data and grep it according to how the diagram below shows the clusters.

leximancer diagram

Most important CPAN Modules

Here, we have an enormous source of data about respondent's favourite CPAN modules. Again this is a rich source of data, and I encourage you to look at the table. The top 6 of 1985 packages or modules mentioned are DBI, Moose, Catalyst, CGI, Data::Dumper, LWP and DBIx::Class indicating strong adoption of Modern Perl by these respondents

CPAN Workflow

These questions were asked on a Likert scale. In the graphs, 2 corresponds to "Strongly Disagree", and 7 to "Strongly agree". 1 corresponds to "Not applicable" Most people always run CPAN tests, but will frequently force install. Many try to understand why tests fail, but reporting test failures is rather patchy.

Cross platform Perl

This graph, (on the same scale as the graphs from the previous section) shows that generally people find cross-platform Perl programing reasonably easy, although this is not an issue for around 600 respondents

Perl Certification

On being asked if Perl certification would be useful "for me" or "for other people" respondents seem to be divided. Most respondents feel it would not be useful for them, but feel that it may be somewhat useful for others. The graphics for this result are here.

Code for this analysis

Almost all of the code for this analysis has been written in R. It's been written so that the survey can be repeated in future years so long as the questions remain reasonably similar. The list of questions (unfortunately not as nicely formatted as I'd like) is here
Last modified: Thu Nov 25 10:48:34 EST 2010