100% I love a cheeky little Lagrange layover myself, but when I was having consistent issues with hangars and landing services it was annoying keeping a running tally of which spots would let me refuel.

Yeah, not saying I want them to do it any different. I'm definitely pro-sim SC, but things like QT fuel capacity are things that can wait til the "beta" phase to be fine tuned.

"but not in the game's current state" is such a necessary addendum to all discussions of balance and sim aspects.

I absolutely love how tedious and involved cargo is now. I don't love losing 20 minutes of loading time to a random piece of discarded landing gear blowing up my ship as a leave hangar.

Yes, that's true. I'm thinking of it in the context of interactive visualizations where you have the user manipulate parameters on a front end. Having those pre-calculated vs stalling out the application.

I was projecting a specific problem I had with a dev at my org to a general problem space lol

That's a hell of a career. You should def be proud of it.

What program are you in?

I second this 100%.

I'll add that (in my experience) SQL is an easier thing to modify and debug when you get used to it.

SQL is more powerful than you realize. As an example, I've applied logistic regressions to entire datasets in PrestoSQL.

(This was an entirely unnecessary thing to do, but just to say what's possible)

Just gotta set your synchronous REST timeouts to 10 minutes

The problem is in the deployment of the tool.

Additional transformations add unnecessary load to whatever device you are trying to represent the data on.

Python and R are powerful, but running the query/transformations in a hosted database and then just passing the raw numbers through is faster.

This! The objective function of a model is just math in isolation, but the moment it is applied in vivo, the objective becomes embedded in the cultural context by which it is used.