Blogs

RedMonk

Skip to content

Programming Language Framework Traction on Hacker News

In February, Bryce Boe – a UCSB Ph.D. student – did an interesting analysis of StackOverflow community data. Specifically, he examined the relative popularity of web frameworks.

It’s a very interesting piece of work, particularly given the market interest in framework performance, but unfortunately Node.js was omitted. Curious to see how it fared, because both RedMonk Analytics and our inbound inquries agree interest in the framework is spiking, I ran some numbers yesterday. Because I didn’t have the latest StackOverflow dataset cut up chronologically yet, I decided to work with the Hacker News dataset we have on hand instead.

For the sake of time, I kept the analysis down to four frameworks: Django, Node, Rails and Spring. There are dozens of others I could have included obviously, and as interest dictates I may add others to this list. Here is how they fared.

Django

The Django trendline shows nothing remarkable. Solid growth, with predictable seasonal fluctuations in discussion.

Node.js

The essential feature of this chart is the growth from, effectively, a standing start. From a tiny project, this has grown into a much discussed and debated platform.

Rails

Rails is periodically dismissed as an overhyped, flavor of the month technology that is in decline. The data here does not support this claim. Rails remains a growth platform, and eclipses the other web frameworks in total discussion volume.

Spring

The Java framework shows steady growth, but the Y axis indicates that its overall discussion volume is modest relative to the competition.

Conclusions

It can be difficult to properly evaluate relative growth rates from visualizations of this type, and unfortunately Google Spreadsheets don’t support trendlines, so I’ve calculated the slopes for the different frameworks as a reasonable proxy for growth rate. Basically, a flat line has a slope of zero, so the higher the slope the greater the growth rate for our sample. With that, the slopes for the selected frameworks are as follows:

  • Django: 12.01
  • Node: 11.27
  • Rails: 19.31
  • Spring: 3.76

What this sample indicates, then, is that Rails is still outperforming competitive frameworks on both growth rate and total discussion volume. What’s interesting is that the much younger Node.js is beginning to approach Django both in growth rate and visibility. Spring, for its part, lags its dynamic language brethren considerably on both fronts.

Remember, of course, that this is an observational sample of Hacker News data, and as such it cannot be considered representative of horizontal developer traction. It is just a datapoint. But given that Hacker News is an aggregate of an engaged, active body of developers we would argue it’s an important datapoint.

Update: Apologies. All of the above graphs were supposed to be interactive, but Google Docs embed scripts were behaving very erractically so I’m serving the static images instead.

Credit: We’d like to thank Ronnie Roller of iHackernews.com for the HN dataset, as well as Jonathan Keith, Steve Morin, and Joseph Turian.

Categories: Programming Languages.

Tags: , , , , , , , ,

  • http://dberkholz.com/ Donnie Berkholz

    Interesting how mentions of Spring peak in mid-winter. Perhaps there’s some seasonal effect for that particular one. =)

  • http://dberkholz.com/ Donnie Berkholz

    I’m curious how you calculated those growth rates. Is it a linear regression or just based on initial and final counts? Are they based on percentage growth rather than absolute numbers, so they’re size-independent?

  • http://redmonk.com/sogrady sogrady

    @Donnie Berkholz: Tableau actually did the calculations this time around, but it’s simple linear regression i believe.

    as for Spring, that is funny ;)

  • http://redmonk.com/sogrady sogrady

    @Donnie Berkholz: oh, and it’s based on the incremental numbers, not start/finish counts.

  • http://www.tom-carden.co.uk Tom Carden

    I’m surprised all the frameworks were growing. Do you have the numbers for baseline Hacker News activity over the same period? Were other common topics (Facebook, Twitter, YC, iPhone, Android, techcrunch, etc.) flat or not?

  • http://sigviper.blogspot.com/ fastviper

    Is the trend corrected for overall grow of Hacker News community? What part of the rise is explained by rise of interest and what part by the framework itself?

    Is the growth of for example Django correlated with growth of Spring? If the answer is yes, coefficients are correlated in model and analysis is totally wrong. In that case, statistician is required to find another variable explaining growth of both frameworks. I’d suggest number of posts per day on HN.

  • http://www.josefrichter.com Josef Richter

    Nice! Maybe some data on sny PHP framework, please?

  • Pingback: pinboard March 25, 2011 — arghh.net