Data visualisation

[Quafadas]

@Dorodri @Kvintochka

Something to think about, if you are interested. The last few lines of this file, set out a sketch for plotting.

https://github.com/Quafadas/scautable/blob/66a3a21874f8a1f4990a123556fb882fb55d627c/examples/src/titanic.scala

I'd be curious to see if it "works" for you, both in terms of actually showing plots and / or thought process. Does it make sense to you? Do think you could invent your own visualisations in this paradigm?

No worries if it's something you don't have time for, but I'd be curious to get any feedback ...

[Dorodri]

Just tried it out—it works out of the box. I believe this serves as a good example not only for scautable but also for dedav4s.

Does it make sense to you? Do think you could invent your own visualisations in this paradigm?

Sure, I found it very straightforward to use. My limitation would be my lack of familiarity with Vega rather than any issue with the libraries.

Overall, I think this is the way to go. My only concern is having to use mechanisms like @implicitNotFound and GetTypeAtName[K, S, V] outside scautable’s internals, but I suppose there isn’t much we can do about that.

[Quafadas]

my lack of familiarity with Vega rather than any issue with the libraries.

Vega is something of an arbitrary choice... although I think it's a good one. It has the advantage of being an old, mature project with lots of prior examples and great documentation... so it's quite "vibe-code" friendly and LLMs are great at explaining what's happening.

I got a lot of help from an LLM when working those examples through

My only concern is having to use mechanisms like @implicitNotFound and GetTypeAtName[K, S, V] outside scautable’s internals, but I suppose there isn’t much we can do about that.

This is... a very good point, to which I don't have a good answer. For the sake of discussion

@implicitNotFound simply provides a better error message if it can't summon the correct implicit. That one is a scala standard method and arguably defensible on those grounds.

GetTypeAtName[K, S, V] and friends is both custom and non-obvious. However, I don't think it's possible to abstract over named tuples without some construct like this.