Plenty of fish in the C, IEEE finds in language popularity contest It's no surprise that C and Java share the top two spots in the IEEE Spectrum's latest Interactive Top Programming Languages survey, but R at number five? That's a surprise. This month's raking from TIOBE put Java at number one and C at number two, while the IEEE reverses those two, and the IEEE doesn't rank assembly as a top- …
If you haven't heard it, it's because you have been avoiding the articles about "enterprise big data" (which is a quite an understandable reaction). Everybody has been packaging R with whatever big data system they are flogging.
It's an open source statistical language, and a GNU project licensed under the GPL. It is replacing the various proprietary statistical packages which people had been using before.
I'm sure that it's a useful thing, but I suspect that the IEEE's ranking has been tilted by all the marketing bumpf being put out by the enterprise vendors (Oracle, Microsoft, etc.) who are rushing to support it recently. I seriously doubt that there's actually more people using R than are using PHP or Javascript.
Once upon a time, when dinosaursmainframes roamed the land there was the proprietary SPSS (Statistical Package for the Social Sciences). SPSS inspired researchers at Bell Labs to produce S to handle statistics. S inspired Gnu to produced a FOSS version, which they called R as T was already a dialect of Scheme.
My wife does most of her work in R these days, as she handles a lot of statistical analysis for government departments and various industries. It's an interesting language, syntactically weird compared to mainstream languages and takes a whole new mindset to use well, but incredibly powerful at what it does, and has the best interfacing to random databases I've seen.
"Once upon a time, when dinosaursmainframes roamed the land there was the proprietary SPSS (Statistical Package for the Social Sciences)."
I'm still paying a butt-load of money annually to use SPSS in our computer labs. But that may be because the current crop of PhD Sociology profs grew-up using SPSS so that's what they are sticking to and don't talk about anything else...
Unlike the dinosaurs, SPSS is still around and IBM (who owns it now) is more than happy to take (a lot of) your money for it every year.
That's because, in much of Blighty, the "R" in R is silent. That's the true reason why Scotland should stay in Britain. Otherwise 7% of the English *language* will disappear on BBC. It will drown in the cold, cold Atlantic. It will be no moa.
I was wondering why my appreciation of "R" was so different from that of the other posters, so I clicked the icon of the programming language whose logo is a large R in a circle (reminiscent of the Ottawa Redblacks football team) to discover that I had mis-remembered its name: "Rust is a systems programming language that runs blazingly fast, prevents segfaults, and guarantees thread safety." Yeah, shoe-ah.
R is an open source variant of S-Plus. I started using it when I received a negative value for a variance from Excel. That was many moons ago now and Excel and most other spreadsheets have gotten much better at such things, BUT Excel (and SPSS, STATA and the like) still costs infinitely more (divide-by-zero error comparing costs) and has vastly less ready functionality. R is used in medicine, climate models, finance, geology, archaeology, paleontology, GIS mapping applications, and good old fashioned statistics. I've talked to recent grads that studied statistics (usually specialized areas) and the teachers are tending more to employing R because of the cost and because writing codes makes the student actually think about what they are doing. It's also handy for things like scraping tables in web sites. I can't see it competing with a more general purpose language though. Why bother?
"R is an open source variant of S-Plus."
That's a bit inaccurate these days. One of the (many) reasons R has skyrocketed up the rankings in the last couple of years is a move away from its original roots. You can pin an enormous amount of that down to one man and his herculean efforts - Hadley Wickham. ggplot2, plyr/dplyr and reshape have transformed R from an esoteric prototyping language into something really accessible and powerful.
Unfortunately for R all those good ideas have been lifted and ported over to Python, which under numpy/pandas looks and feels the same while also being a full, robust programming environment.
But still, nice to see it up there.
Tiobe puts R at 17, with a 1.5% ranking. Go is not even in the top 50 and is lumped in with languages with too small of a market share to measure meaningfully.
A more "diverse" source of data such as the IEEE uses isn't necessarily a better one. Some of the sources which the IEEE uses are subject to being tilted by marketing drives, or by HR driven laundry list job ads or CVs, where whatever is "cool" this month gets spammed over the Internet whether it's relevant or not.
I find it extremely implausible to suggest that R is more commonly used or in more demand than PHP, Javascript, or C#. R has a very narrow use case, while the others I mentioned are very widely used. The IEEE ranking seems a bit lacking in usefulness.
"D ranks as 6th which is substantially higher than Ladder Logic"
Which when you think about it, is utterly implausible. Almost every factory in existence runs on ladder logic, and most individual machines have unique custom ladder logic programs. That's a lot of ladder logic being written, and re-written every single day around the world.
However, people who program industrial systems have their own forums and web sites, and don't generally frequent the same ones that the IT business does. It's two separate works, and seldom do the two meet (I know because I have a foot in both worlds). The IEEE doesn't look at the ladder logic forums, and so doesn't see much of that world.
Yeah, I agree. It is implausible.
To be honest, this whole exercise is implausible. How do you even define a sensible measure of "popular"? Searches on the web won't do it - that would favour new upcoming languages with a lot of learners. Should it mean lines of code in production, lines of code in active maintenance, or number of person-hours of development?
Etc.
That is the only way to express this otherwise subjective list of "popularity;" Digging through the various code repos online. Discussion forums can also have some useful data, but let's be honest; most of what is and what is not written in code is not available online. There are many, many, many petabytes of storage and code within there that is well behind firewalls and not going to influence this silly table.
The language you use to code with is there because your site/company approves of its' use, you know it, the guy who left and wrote tons of your biz logic and other automations are written with it, it's a known thing. Sometimes it's a legacy thing and everyone either agrees with that and accepts it, or there is a push of some type to get into a more modern framework and what have you.
Most everything coded I see is in Perl and Python and shells, since that is where I do my thing as a devops admin who knows automations, but not full application building and other larger software ecosystems. I took up C99 when I needed to pick coding back up, but I rarely use it since I'm just writing simple automations to keep the systems doing whatever it is they need to do. Is that less popular with me? No, that's silly. C99 is plenty important, but it's not the language I reach for when I can get the same functionality out of Perl, with less effort. Or shell, when you don't need any fancy logic in it. If I were a hardware designer or app guy, then I'd be into the lower level languages for the obvious reasons. I guess this is where I say popularity is subjective, especially in a context of what is and what is not appropriate for your project or product. You run into trouble when you try and shoehorn an unsuitable language for a project because the syntax is pretty, or other subjective silliness that any normal enterprise would naturally avoid.
I find Python to be pretty and effective, but I'm new to it, so I do things in Perl until I get comfy with Python. I'll get paid either way, so learning Python to me is adding value to my config, but not a requirement unless my site is written in it.
HTML definitely isn't a programming language: it's an SGML dialect.
SQL, as you rightly point out, is a Turing-complete programming language even if I think the chosen semantics are extremely unsuited to the domain.
Getting these two wrong tells us a lot about this language beauty contest.
~ Never heard of R before, need to brush up maybe...
~ Have a suspicion that VBA would be higher than 18 if VBA was fully taken into account. i.e. every other middle manager's favorite XLS/XLSX, and half the banks around the world etc...
~ It'd be interesting to go further in that breakdown too and discover how Componentized-C# (Unity3D) compares with Blueprints/C++ (UE4/Unreal Engine)
Whilst I wouldn't really consider HTML to be a programming language, if you're going to have it in there and also have a "Web" category, then surely it's got to score top in that category?
Secondly, SQL below Assembly! Seriously? There's an element of fantasy programming that seems to sneak in with some pretty obscure but very fan laden languages scoring way over these half century old workhorses. CUDA over SQL for instance? Seriously how many people even use that at all rather than just say that they use it because it seems cool?
Agree. Python at #3 is cool, that being my primary language these days.
However... if you look at, and analyze, job postings, there is no way it rates above C++ and C#. What is happening instead is that it is an incredibly powerful Swiss army knife that can assist any developer or sysadmin, so it often gets added as a desirable trait.
So the posting instead typically looks like...
Need to have
Go - 8 years experience
Windows 10 - 5+ years system experience
....
Nice to have
Python or Ruby scripting skills
Or you may also find it on a QA job because their scripts are using it. Core, line of business, programming in Python, outside of web apps? Not, quite, at #3.
JavaScript and its sundry frameworks ... a typeless brace-adorned functional stateful mutant gibbering monster from the festering depths of twen-cen Internet-bubble-powered clueless hackery. A fatwa should be issued against it. Supposedly people who are not CS engineers able to keep suicidal thoughts and rage attacks under control via daily meditation are meant to write fit-for-purpose programs therein? Delusion in the 21st century has reached new levels. Then a Node.Js grogan is deposed on your tabletop...
I have recently discovered elm. It is in the ML lineage. There is theory and philosophy behind it. Beautiful. Plus it transpiles to JavaScript, so portability is guaranteed.
It's time to have fun again!
One of the great things about R is that plays nicely with 'real' languages like C. Manage your data ingestion and basic manipulation in R, then when you need some heavy computation, like Monte Carlo simulation, pass the load onto a C routine. Then display the results graphically in R.
Another feature of R is that it itself has been wrapped into 'reproducible research', where R code is written in-line in text that is destined for human consumption, coded in markdown. Write the script, analyses and present the results, then 'compile' the report to produce HTML or PDF output. All the steps in data ingestion, cleaning and manipulation are embedded in the script so others can read the script and run it themselves. That's a great step forward for honesty in data analysis.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
