Thread by @yestiseye

@BenMoretti @KarlAlexPauls Yes, but at this stage primarily for the graphing capabilities. Any data manipulation (which is still modest in #whtr…that'll change) is handled with python numpy. The internals of R don't like me throwing that much data at it in one go, but I'm only using the basic i/f. 1/x

@BenMoretti @KarlAlexPauls So the first step in the total rewrite of #whtr is an actual system architecture. Of which maybe a C bootstrap into low level R is called for. I'm also investigating Julia as an option.
But Spark looks interesting. esp since the 550W power unit on my home server failed. 2/x

@BenMoretti @KarlAlexPauls I'm look at them right now & I can see the bitcoin effect. Wondering if I can justify the need for 800+W.
Probably not. But the motherboard is nearly 2 decades old. Upgrade time I think. It is a rather large case, new m/b needs to fit right. You are clearly a bad influence.😉 3/x

@BenMoretti @KarlAlexPauls I do love browsing for hardware, with the emphasis on the *hard* stuff🤨 It's my shoe shopping.
We should collab one day @BenMoretti …should you ever want to rip a hole in the spacetime fabric (you're welcome too @KarlAlexPauls 😁)
anyway, the CPUs have been busy… 4/x

@BenMoretti @KarlAlexPauls I must admit I expected a more dominant display out of the non‑hydro renewables… this is strictly preliminary, needs yet to pass a smell test. (I did do 3 month batches) 5/x

@BenMoretti @KarlAlexPauls Clearly a strong winter/summer divide in the data. Note that the solar *is* there - but with so much data, very easy for it to get swamped when interspersed between other data.
Will have more this evening. Reading Snowy 2.0 & *this* 👇 6/6