The State of Ruby and Testing
22ideastreet.com
The State of Ruby and Testing
1–10 of 11 posts
Re: The State of Ruby and Testing
#2Cool survey. I can't help but feel like pie charts are a terrible choice for portraying the information, because they imply that a respondent falls into one (and only one) category, and that all the slices add up to 100%. But I have a sneaking suspicion at least some of the people who have used 1.9.2 have also used 1.8.7 at some point. It'd be much more useful to hear the 90% of rubyists have tried 1.8.7, 85% have tried 1.9.2, etc.
Still, interesting data. I have learned that I am apparently a completely typical ruby user.
Re: The State of Ruby and Testing
#3The pie charts don't really make any sense. For example, a single project can use Cucumber, Capybara, and RSpec, or none of them, or any combination. But in the article and its graphs they are presented as mutually exclusive options.
Separating the results into exclusive types (testing framework, browser integration framework, assertion macros, etc) would make a lot more sense. Or asking what tool is being used for a particular task (unit testing, integration testing, mocking/stubbing, assertion macros, etc).
Re: The State of Ruby and Testing
#4Thanks rauljara and automach. I agree that the pie charts were probably not the best way to represent the data that was gathered. The questions could all take multiple answers, and I just summed the results to figure out how big of a piece of the pie each section would get. I made a quick edit to the post to clarify this. Thanks for taking the time to write that up, would appreciate any tips on how to better represent multiple answers per person per question.
Re: The State of Ruby and Testing
#5The pie charts don't really make any sense. For example, a single project can use Cucumber, Capybara, and RSpec, or none of them, or any combination. But in the article and its graphs they are presented as mutually exclusive options. Separating the results into exclusive types (testing framework, browser integration framework, assertion macros, etc) would make a lot more sense. Or asking what tool is being used for a…
So for all projects that use a combination of tools (the vast majority of projects I would guess), the collected data is invalid. That's too bad.
Re: The State of Ruby and Testing
#6Interesting survey but with some flaws. Ruby seems to be reduced to "Rails" and "web". While this certainly is the majority of code being written, it's still by far not the only code. Testing a library with capybara or selenium is just not possible. What about components and projects not using rails such as stand-alone daemons not even speaking http?
Re: The State of Ruby and Testing
#7Thanks rauljara and automach. I agree that the pie charts were probably not the best way to represent the data that was gathered. The questions could all take multiple answers, and I just summed the results to figure out how big of a piece of the pie each section would get. I made a quick edit to the post to clarify this. Thanks for taking the time to write that up, would appreciate any tips on how to better represen…
The colors are very hard to differentiate with the larger datasets, such as the 2 testing frameworks charts.
Re: The State of Ruby and Testing
#8The pie charts don't really make any sense. For example, a single project can use Cucumber, Capybara, and RSpec, or none of them, or any combination. But in the article and its graphs they are presented as mutually exclusive options. Separating the results into exclusive types (testing framework, browser integration framework, assertion macros, etc) would make a lot more sense. Or asking what tool is being used for a…
So for all projects that use a combination of tools (the vast majority of projects I would guess), the collected data is invalid. That's too bad.
It might not make numerical sense, but it still communicates its intention.
Re: The State of Ruby and Testing
#9Thanks rauljara and automach. I agree that the pie charts were probably not the best way to represent the data that was gathered. The questions could all take multiple answers, and I just summed the results to figure out how big of a piece of the pie each section would get. I made a quick edit to the post to clarify this. Thanks for taking the time to write that up, would appreciate any tips on how to better represen…
The colors are very hard to differentiate with the larger datasets, such as the 2 testing frameworks charts.
Doesn't really matter since the legend is ranked by slice share and the slices get smaller clockwise.
Re: The State of Ruby and Testing
#10Thanks rauljara and automach. I agree that the pie charts were probably not the best way to represent the data that was gathered. The questions could all take multiple answers, and I just summed the results to figure out how big of a piece of the pie each section would get. I made a quick edit to the post to clarify this. Thanks for taking the time to write that up, would appreciate any tips on how to better represen…
The colors are very hard to differentiate with the larger datasets, such as the 2 testing frameworks charts.
True. I would like to have used a nice JS library or something so you could drill down better. Putting numbers on the charts would have been nice to get a sense of the actual sizes (especially on the ones with very small slivers.) Will keep this in mind for next time!