I've got a small hack purpose-built to reduce some...

# coroutines

[jimn]

I've got a small hack purpose-built to reduce some data group-by and pivot headaches working with pandas, written and re-written with zero-copy in mind for large memory mapped fixed-width-format. it's one source file + tests and project trappings.
There are two things I have not yet accomplished with this a) researching the best idioms for parallelizing the working Flow code, and b) acheiving fully orthogonal rows and columns manipulations with one class. index operators work, and row iterators work, but indexed row iterators simply inherit some other agenda. this started as delegation, and has been hammered on but not entirely made monadic.
I'd love any suggestions or PR that might further a+b along
Thanks!
https://github.com/jnorthrup/columnar/blob/master/src/main/java/com/fnreport/mapper/Columnar.kt#L106