Programming Language Popularity Survey

An anonymous reader writes "David N. Welton yesterday posted a study of the Programming Language Popularity. Is SQL your fave, or perhaps you're interested in the 'Click Price of PHP' or 'Craig's List Jobs'? Needless to say, my favorite languages (Prolog and Common Lisp) did not so much as register on the survey."

Comments

Not too valid.

[BigZaphod]

This "study" has been showing up all over the net over the past couple days. I don't understand why it keeps getting so much attention. It is basically just a bunch of google searches and pretty graphs which tell you very little. It claims to be a programming language survey and yet has entries like "Windows programming." What language is that? Heck, there isn't even a c++ option and c++ is probably still one of the most popular of all languages.

... Not getting the point...

[DavidNWelton]

Unfortunately, I don't seem to have communicated the idea behind the survey very well. Let's give it a go here before too many snide, uninformed comments show up about "windows programming".

*) It is for fun. If I were investing the time and money to produce a survey for you to base your business on, I would not give it away for free, or I would have at the very least aimed for publishing it in a magazine like Dr. Dobbs.

*) This means that I used the resources at my disposition as best I could. Those include freely available sources on the web. Part of why I think the survey does have some broad validity is that I tried to find a variety of sources (which you would realize if, you uhm, actually read the article). In a future version, I think I will also attempt to include data from Amazon about books available for whatever language.

*) Why isn't XYZ in the list?! There are lots of programming languages out there. In a recent gig, I was programming Erlang, and liked it a lot. But to give some sort of cutoff, I chose the Overture dollars/click data, which isn't present for lots of "minor" languages. By the way, Cobol figures better in Overture than Lisp and Prolog do, even though Lisp is in my opinion far, far more interesting.

*) If you think the methodology could use improvement, well then by all means send me some email with your ideas, or if you're the independent sort, go off and do your own work if you think you can do better.

*) Google Hits. Yes, I used that. I also used 3 other data sources, so RTFA before you make uninformed comments. In any case, even if there are some problems with Google hits, they *do* represent the visibility of the language. Suggestions on how to deal with specific queries such as VB vs "visual basic" are of course welcome.

*) "Windows" and "Unix" programming. Those who engage their brains for a second or two might come to the realization that, no, they are not programming languages, but queries I threw in as extra data points, for the fun of it. Sheesh.

Does that put it in a clearer like for those of you with your knickers in a knot?

Thanks,
-Dave

Can't Find the F-Word (i.e. FORTRAN)

[reporter]

A quick scan of the report indicates that the F-word (i.e. FORTRAN) cannot be found. We are finally ridding ourselves of this albatross.

Don't get me wrong. I program in FORTRAN for a living, but I have compared it to C and Java. FORTRAN is an ugly language. It fosters the same disgust that x86 assembly code would foster if we lacked compilers and were forced to program in it on a routine basis. FORTRAN is just a bunch of mishmash that has grown to include every interesting feature that catches the fancy of programmers. The current definition of FORTRAN even includes pointers!

Re:Can't Find the F-Word (i.e. FORTRAN)

[beliavsky]

Since the 1990 standard it's spelled "Fortran". I've noticed that people who misspell it often have little knowledge of it.

In their handling of arrays, C and C++ are closer to assembly language than Fortran. Allocating a 3-D array in Fortran is done with just

real, allocatable :: x(:,:,:)
allocate(x(2,3,4))

In C/C++ you declare a pointer to a pointer to a pointer -- talk about ugly!

Fortran 90/95 pointers are less powerful (and dangerous) than those of C/C++ but are still useful. Often I find myself referring repeatedly to an array section. Using a pointer I can refer to this section more concisely.

A few methodology problems

[AuMatar]

Beyond the ones listed already:

*For things like TTCL, shell, and SQL, these are usually secondary skills wanted along with another language (C, perl, PHP, Java). This means an artificially inflated count for them

*Bias in the web. A lot of programming subareas just don't have much web presence- firmware for example. A lot of these are tilted to C, Cobol, and Fortran. Nobody writes firmware in Java.

*Internal code. Most projects are never released to the public. Unless they have a job opening being advertised, we don't know what language they're using.

*Job listings- there's an inherent assumption that web job listings are reepresentative of the industry as a whole. It may be, but I have no evidence either way. It wouldn't surprise me to see web-realted jobs have a higher proportional representation.