This change in ranking would deserve a separate post in this forum in my opinion, for better publicity (I came to post it on the forum yesterday, but there was already a discussion started here).
I guess mine is a very personal perspective. The problems I had/have very rapidly need awkward vectorization syntaxes to remain useful, and using numpy quickly becomes unnatural (though with numba and stuff it is possible that I could have a different experience today). I guess comparing numpy to Fortran makes total sense if the alternative is C, or C++.
(I have no experience whatsoever, but Nim is very interesting from what I could see).
Yes, so speed is the issue. Absolutely, I have the same experience and that is why I like Fortran.
TIOBE is still not too relevant, but just for fun: In August 2021, Fortran ranks #13: index | TIOBE - The Software Quality Company
For reference, it was #20 in April 2021, #50 in July 2020 (lowest ever), and #10 in March 2002 (highest ever).
Itâs very relevant as long as Fortran keeps rising. It will become irrelevant once we start falling.
The text says:
Nowadays we have the same with data mining and AI. Programming languages in these fields are booming. The most striking example is Python that took over the second position from Java. Even old languages see a revival because of this, like the surge of Fortran.
What do you think of that explanation? Data mining and AI are trendy for many years. Can it explain the rise of Fortran in one year?
The books and blogs on data science and machine learning Iâve seen mostly use Python and R. Fortran is rarely mentioned. Those domains do rely on fast linear algebra, so Fortran is not a bad fit.
I saw that. I doubt thatâs the explanation. I think itâs our efforts here that helped it propel upwards. The graph at TIOBE Index - TIOBE is also interesting, as mentioned above. Here is a screenshot for future reference.
The previous peaks correspond to the release of the Fortran Standard in 2003, 2008, 2018. The initial decine from 2001 I would say is characteristic of my perception of the general lost of interest in Fortran. And finally the last peak in 2021 is our effort. It feels to me it is still at the level of noise, but if we can get higher than in 2001, then I think we are onto something.
Hopefully, people are also starting to get seasick from the C++ *,&,&&,⊠stutter in its attempt to leave behind the fixation with memory addresses. I see C++ is trialling Intent(In): https://cppx.godbolt.org/
I also do. Itâs a problem: there is an automatic algorithm, then people are just giving their opinion, they have found an easy explanation based on no facts. They are not investigating. But they have a large audience. And other sites just relay the same opinionsâŠ
I wonder why Fortran is not listed in the most popular Programming, scripting, and markup languages section of the 2021 Stack Overflow developer survey. Is it really less popular among survey respondents than APL and COBOL, or was there no option for respondents to list it?
JavaScript | 64.96% |
---|---|
HTML/CSS | 56.07% |
Python | 48.24% |
SQL | 47.08% |
Java | 35.35% |
Node.js | 33.91% |
TypeScript | 30.19% |
C# | 27.86% |
Bash/Shell | 27.13% |
C++ | 24.31% |
PHP | 21.98% |
C | 21.01% |
PowerShell | 10.75% |
Go | 9.55% |
Kotlin | 8.32% |
Rust | 7.03% |
Ruby | 6.75% |
Dart | 6.02% |
Assembly | 5.61% |
Swift | 5.1% |
R | 5.07% |
VBA | 4.66% |
Matlab | 4.66% |
Groovy | 3.01% |
Objective-C | 2.8% |
Scala | 2.6% |
Perl | 2.46% |
Haskell | 2.12% |
Delphi | 2.1% |
Clojure | 1.88% |
Elixir | 1.74% |
LISP | 1.33% |
Julia | 1.29% |
F# | 0.97% |
Erlang | 0.79% |
APL | 0.65% |
Crystal | 0.56% |
COBOL | 0.53% |
LOL, the analysis and the plot looks like some cryptocurrency or stock market price prediction
Does it somehow related with the fact that Intel seems decided to make their Fortran and MPI free, now branded as Intel OneAPI?
I feel the free Intel OneAPI toolkits may have some significant impact on Fortranâs future.
at least for Windows users, if people can get free high performance intel Fortran + MPI which tightly integrated with the most popular IDE the visual studio, this can make Fortran much more accessible.
Good point, it might very well be.
In the September TIOBE index, Fortran is 17th, between Matlab and R:
https://www.tiobe.com/tiobe-index/
Down from 13th in August.
There are various plausible measures of programming language popularity. Fortran ranks much lower based on GitHub popularity. The GitHub top 10 are
- 2,003,583 JavaScript
- 1,493,928 Python
- 1,095,520 Java
- 666,343 HTML
- 522,005 C++
- 506,754 PHP
- 430,664 C#
- 374,092 C
- 353,038 Jupyter Notebook
- 349,520 Ruby
Searching language:Fortran
on GitHub gives 17,453 results on GitHub. According to GitHut Fortran ranked 39th. R, Julia, and Matlab rank 32nd, 33rd and 37th. According to RedMonk, which uses GitHub and StackOverflow popularity, Fortran is not in the top 20.
The Fortran data used for GitHutâs statistic seems to be incomplete for some reason:
There are a couple of languages that jump into existence at the same point. A similar thing can be seen with Julia, which was created in 2012 and therefore shouldnât start from zero at 2014 (time is difficult to tell because the x axis has no label, but GitHutâs data starts at 2014).
TIOBE seems to favor old programming languages. Last month, Fortran appears to be more popular than R on the TIOBE list, which I find hard to believe.
Really the TIOBE list is just really bad. A significant portion of it comes from websites like ebay and amazon listings because those are clearly both relevant information. Each language also has a subjectively chosen number for how to weight itâs results. Metrics like PYPL are much better if you actually care about popularity.
So what? It doesnât determines if a specific task x
is resolved any better .or. worse than earlier (or by other languages).
By the same token, the surveyorsâ perspective assigns Fortran thrice as much popularity now than a year ago. «Back then», there wasnât even an explicit comparison of their rating with the one assigned the year earlier. (One probably could establish this with some archaeology on archive.org):
| month | rank | rating | change | Fortran trails | Fortran just ahead of |
|---------+------+--------+--------+------------------------------+------------------------------|
| 2020/01 | 34 | 0.313% | NA | ABAP (0.364%) | Kotlin (0.294%) |
| 2020/02 | 34 | 0.229% | NA | Lua (0.235%) | PowerShell (0.209%) |
| 2020/03 | 33 | 0.29% | NA | F# (0.32%) | Lua (0.28%) |
|---------+------+--------+--------+------------------------------+------------------------------|
| 2020/04 | 34 | 0.33% | NA | F# (0.34%) | Lua (0.33%) |
| 2020/05 | 34 | 0.32% | NA | Ada (0.33%) | PowerShell (0.32%) |
| 2020/06 | 37 | 0.27% | NA | OpenEdge ABL (0.29%) | Lua (0.26%) |
|---------+------+--------+--------+------------------------------+------------------------------|
| 2020/07 | 50 | 0.22% | NA | C Shell (0.22%) | #51 -- #100 |
| 2020/08 | 42 | 0.31% | NA | Lua (0.34%) | Haskel (0.31%) |
| 2020/09 | 37 | 0.36% | NA | OpenEdge ABL (0.38%) | Ada (0.34%) |
|---------+------+--------+--------+------------------------------+------------------------------|
| 2020/10 | 37 | 0.38% | NA | COBOL (0.38%) | Lua (0.35%) |
| 2020/11 | 34 | 0.41% | NA | ABAP (0.43%) | LSIP (0.41%) |
| 2020/12 | 32 | 0.43% | NA | Dart (0.45%) | COBOL (0.43%) |
|---------+------+--------+--------+------------------------------+------------------------------|
| 2021/01 | 30 | 0.46% | NA | Prolog (0.48%) | COBOL (0.42%) |
| 2021/02 | 23 | 0.63% | NA | SAS (0.66%) | D (0.59%) |
| 2021/03 | 22 | 0.83% | NA | SAS (0.87%) | Scratch (0.78%) |
|---------+------+--------+--------+------------------------------+------------------------------|
| 2021/04 | 20 | 0.91% | +0.58% | Matlab (0.99%) | SAS (0.89%) |
| 2021/05 | 20 | 0.83% | +0.51% | Perl (1.04%) | Objective C (0.79%) |
| 2021/06 | 17 | 1.07% | +0.80% | Swift (1.10%) | Delphi/Object Pascal (1.06%) |
|---------+------+--------+--------+------------------------------+------------------------------|
| 2021/07 | 14 | 1.12% | +0.90% | Go (1.17%) | Groovy (1.09%) |
| 2021/08 | 13 | 1.14% | +0.83% | Classic Visual Basic (1.23%) | R (1.05%) |
| 2021/09 | 17 | 1.01% | +0.65% | Matlab (1.02%) | R (0.98%) |