I looked at bit at the github activity for various languages in terms of repos created in the past 2 months using Github advanced search.
language | number of repos created in the past 2 months |
---|---|
Java | 371K |
TypeScript | 165K |
C# | 147K |
C++ | 102K |
C | 64K |
Kotlin | 50K |
Dart | 44K |
Go | 38K |
Swift | 29K |
Rust | 15K |
Scala | 3K |
Pascal | 2K |
Haskell | 2K |
Objective-C | 2K |
Julia | 2K |
Fortran | 657 |
Haxe | 552 |
F# | 520 |
OCaml | 494 |
Prolog | 368 |
Zig | 245 |
Erlang | 220 |
Nim | 217 |
D | 185 |
Crystal | 119 |
Vala | 61 |
Pony | 11 |
Chapel | 1 |
My interpretation:
Languages that are used by students and academics appear to have a boost on Github (Fortran, OCaml, Haskell, Java). To get over 1K you seem to need to have a unique domain where are you are or have been dominating (except for Scala, which is an outlier).
I didn't expect C++ to have so many new repos, and expected the distance between C# and C++ to be bigger. Is this because the C++ standard library is weak, or do C++ programmers have a stronger inclination to share their projects in progress on Github? I don't know, maybe C# programmers have other platforms they prefer.
Dart being so close to Kotlin and Swift is also a bit surprising; it suggests that people have a strong preference for portability. I actually thought that Haxe was fading away, but it appears to be quite strong for a language that does not receive much exposure. Again, could suggest that portability is a strong factor.
Nim and D appears to be fairly stable, so no real surprise there, but Zig is getting more attention than I would expect at this point in time. I personally don't view Zig as a particularly interesting language design, so does the apparent growth reflect early enthusiasm or does it reflect actual productivity?
Pony and Chapel are interesting languages, but maybe concurrency is not as important to most programmers as we might like to think it is. Erlang has had a lot of exposure, yet it does not have more activity than languages that receive very little exposure. So again, maybe concurrency is not a strong factor when people select programming languages (yet).
(Note: github results vary by a few % for each search, so it isn't precise).