This is a solution written in Red language. It’s a one-liner but split into commented multiple lines for easier understanding. As a bonus, I added decompression also.

Red scripts work via rebolek/red docker hub image. Makefile is fixed so all red scripts get executer with Nix command:

nix-shell shell.nix --run "make red-docker" 

I collected and tested all Red solutions, from other PR-s and Red gitter chat. I hope it's OK, but I wanted to see all Red solutions work.

It seems python3 doesn't run scripts with

python3 src/python/*.py 

like it's written in the make file. Maybe there is a more elegant solution but this works

find src/python/ -name "*.py" -print0 | xargs -0 -n 2 python3

I've tried my luck golfing also in python :)

in 108 bytes: s=lambda s:reduce(lambda a,b:([a[0]+str(a[1])+b[0],1],[a[0],a[1]+1])[b[0]==b[1]],zip(s,s[1:]+'0'),['',1])[0]

Example built for Windows using FASM. https://flatassembler.net/ Uses: FASM macro to invoke C functions C runtime library for "printf" function. Windows kernel library for "ExitProcess"

Added my el cheapo Python solution to benchmarks.

➜ python src/python/benchmarks.py 
Brodul 0.0321615369994106
Luka faster 0.027532846999747562
Luka 0.02394970400018792
Oto 0.03818903700084775
Urban, iterative 0.04462564499954169
Urban, recursion 0.04826400199999625
romunov, babysteps 0.013087426999845775

In the majority of real life scenarios I'd likely just use a more simple solution, such as the solution @robcole presented earlier, but this is over 4x faster and uses a Lexer pattern to seek through the string, so it is interesting and could be a better pattern to use in real life, but practical scenarios (seeking through strings at a framework level, for instance)

Definitely an interesting solution, and while the solution doesn't look the most elegant, this pattern would look a lot more coarse in other languages and that makes the outcome interesting to me!

My attempt at solving the puzzle using three basic R functions (#24) strsplit, rle and paste (paste0 is just a synonym with a default value for a separator set to "").

In addition to the solution I've also added myself to the authors, bumped stats and tried to add R to the run chain. I'm pretty sure I somehow botched the last part.

In the example, There is only 1 C that gets compressed into 1C. If we take the knowledge that all "multi character" will have a number, and all singles do not have this prefix number, we can improve the compression ratio of this problem

It would make sense to run a common test suite against all the solutions through stdin or something like that. Either that or make the solutions also test the empty case (compress("") == ""). I've ran into solutions where the empty case resulted in compress("") == "0" while prototyping, and I wouldn't be surprised if some existing or future submissions had the same issue.

In what other languages would you like to see solutions, and do you know anyone who could provide them?

I would be interested to see Kotlin solution. There are probably some Android devs that use it. And Rust, and Julia ...