Comparisons of dask and bend

Recently came across GitHub - HigherOrderCO/Bend: A massively parallel, high-level programming language.

Whenever, I come across a new parallel library i’m curious to learn more about it and I often want to compare to one that i’m familiar with (dask).

Curious if anyone has thoughts on this and has taken a deeper look beyond reading the README.

May be worth expanding Comparison to Spark — Dask documentation to have a section on comparisons to other libraries e.g. spark, ray, beam, joblib, multiprocessing, concurrent.futures, cloud functions (e.g. lambda), SQL and now bend. I feel there’s lots of documentation/discussions floating around on comparisons with spark, ray and other parallel frameworks. I know it’s hard to be unbiased in these situations but a summary paragraph may help and eventually may help with things like code translation e.g. I want to translate my spark code to dask code. Ultimately it may help with the dask roadmap.

Took a quick look Ray Bell - Dask vs. Bend but mostly spent my time on installation and getting my head around the language

Hi @raybellwaves, thanks for sharing this!

I’ve only took a quick read on the first documentation pages, but I admit I stopped fast because I quickly understood that Bend was not real Python code, and introduced other things like Object and new syntax. Moreover, it is limited on single node scaling.

I think Dask and Bend cannot really be compared, as they don’t have the same target.