Ruby Survey Results

Highlights: 1,000 responders, of which 470 Rails and 530 Ruby; TextMate is the most popular tool globally, but the Vi family is slightly preferred on the Ruby side. Thanks, everyone, for pitching in!

Methodology · The questions were:

1. Is your work mostly Rails, or do you do significant non-Rails Ruby development?

2. Which developer tools do you use? Pick as many as you want.

On the latter, the options were (presented in random order to each participant):

• TextMate

• Emacs family

• Vi family

• Eclipse

• NetBeans

• IntelliJ

• Other (please say which)

I announced the survey on this blog, on the ruby-lang mailing list and IRC channel, and asked someone to do the same in the Rails equivalents. I cut off the survey when the sample size reached 1,000, after six days.

I downloaded a .xls of the results from SurveyMonkey, and did a little hand-sanitation. Lots of people had listed tools (irb, rake, autotest, you name it) that weren’t actually editors. Also, there were a few editors that showed up often enough under “Other” that probably they should have been among the choices.

So I did three flavors of cleanup: First, where they’d listed Aptana and/or RadRails, but hadn’t listed Eclipse, changed that to an Eclipse vote. Second, I regularized the names of Komodo, jEdit, sciTE, and so on. Third, I removed all the non-editor tools.

Then a ran a little Ruby script to group the totals by into the “mostly-Rails” and “significant non-Rails Ruby” buckets, using the labels “Rails” and “Ruby”.

Results · [Corrected: Editing error had undercounted sciTE.]

First of all, the sampling methodology is laughably unscientific; the audience is self-selecting. I asked SurveyMonkey to accept only one result from any given IP address, but we know how much that’s worth. I find the results interesting, but they’d be bounced out of any refereed journal.

You can look at SurveyMonkey’s decent presentation of the results; it doesn’t group by Ruby/Rails, but it does allow you to look at the raw “Other” comments.

Lots of people selected multiple tools; in aggregate, the 1,000 responses named 1,575.

Here are three quick graphs summarizing the results, with bars for any software that got at least 10 votes, which is 0.1%.

Here’s pretty much all the data, in a sortable table (click on the column headers).