Pietro Vertechi

OK! Let's see if other people have other ideas, otherwise I can start integrating it in the docs.

Why not use the IterableTables/Query machinery? If you do: ```julia using CSVFiles, FileIO itr = load(filename) ``` you get an IterableTables of NamedTuples (lazy, you haven't read the data yet)....

My bad, I somehow assumed that CSVFiles would read line by line to produce the `NamedTuple` iterator but apparently not, it tries to read everything at once. Still, it be...

Yes, it's waiting for https://github.com/JuliaPlots/Makie.jl/issues/926 to handle it without too many contorsions, but it's good to keep in mind that axes linking should also take this into account.

You can probably hack your way around it with something like ```julia @recipe(FlippedBarPlot) do scene return default_theme(scene, BarPlot) end Makie.convert_arguments(::Type{ axis is preserved in the recipe `FlippedBarPlot`, so everything should...

Yes, that is correct. You can actually already do ```julia CSV.read(fn, StructArray∘Tables.columntable) ``` which is probably close to optimal (the data structure backing a `StructArray` is a named tuple of...

> `CSV.read` is pretty well documented that you need to pass a valid sink function; Makes sense. I can probably still define `materializer` for a `StructArray` type, so that users...

Would it be possible to define iteration on a distributed table? This way all the tools that only use iteration (like for example TableView) would work on a distributed table.

> Not having row indexing renders this pretty much unusable for a lot of Pandas-type operations on large data - I was super, super excited to find this as a...

I think somehow there should be a trait on recipes that tells you the orientation. Alternatively, this could be done upstream, and AbstractPlotting could support working with named arrays (AoG...