That is interesting! I knew Python was hot, but I didn’t know it was going to be this at all. I haven’t looked through the methodology in great detail yet. Too many words without pictures, so I got bored. But, it does seem a bit different than StackOverflow’s results from their survey: https://insights.stackoverflow.com/survey/2018/#most-popular-technologies
Aww … I was ready to dust off my old hard drive.
I believe stack overflow used a survey, right? While the posted article tried to analyze trends from a bunch of different sources. I would say that asking people directly would provide a better insight into these things, especially as far as something like “popularity” goes, but I think the other data is interesting as well.
My problem with trends is that it can skew different ways depending on other factors. For example a less used programming language may be seen more popular because its also a more difficult language to learn. As a more difficult language, you’re going to have people searching for it more. Then again what’s surprising is that Python is at the top of their lists - not a language I would necessarily call difficult. Of course they’re saying HTML is #9 in web rankings which seems a little odd to me, but again, probably a consequence of that methodology.
Does anyone know why Python is picking up steam all of a sudden? For example Rails put some momentum behind Ruby for a while. Is there something similar behind the Python popularity? Is it big data? And what about big data lends itself to Python? Wouldn’t you want something more low level/performant for that kind of stuff?
It’s the default/easy language binding for “big data” or data science environments like Spark and Databricks, and the default/easy binding for machine learning technologies like Turi Create and TensorFlow.
Performance-wise, the Python bindings are either FFI wrappers over C/C++, or serve as metaprogramming libraries that do codegen for distributed systems.
By metaprogramming, I mean that your sequence of
reducees usually aren’t operating on your actual datasets, they’re just serializing a description of what your Python script did, then running that code in a different language on other remote machines, then giving you a promise back that lets you do more meta-work on the results. But it looks like normal code more-or-less.
That makes sense. Thanks!
I wish I learned Python instead of smelly R.
Another reason for Python’s increasing popularity may be seen in R’s small decline. R peaked at No. 5 in 2016, dropped to No. 6 last year, and is now in seventh place.
Additionally, Python is embedded in several softwares that cater to 3D modelling, and is used as a scripting language to interface with it (for instance, generate different geometries that don’t necessarily come prepackaged for you).
I personally sketch ideas in python, it’s just easier to do and I find it discouraging to write verbose code if I’m still struggling with an idea. I eventually translate/port it to Java/GLSL for real-time speed.
It’s never too late!
It’s a little late as I have other priorities at the moment. Maybe someday.
If someday comes and you needed something Google couldn’t provide, you can drop me a line.