Andy Ferris

Regarding R's `tapply` if you want to apply `fun` to each group you can do `fun.(group(g, x))` (or sometimes `fun.(groupview(g, x))` might be faster/less memory hungry, and there is always...

I'm not sure what you are seeking? Is it this? ```julia julia> g = [1, 1, 2, 2] 4-element Array{Int64,1}: 1 1 2 2 julia> x = [5, 6, 7,...

Yes. Take a look at e.g. https://discourse.julialang.org/t/splitapplycombine-jl-group-enhancements-reaches-version-1-0-0/32675 In your example we can do something like `length.(group(iseven, 1:5))`. You can convert the result to a `Dict` straightforwardly if that works better...

Yes the documentation could definitely be improved. In fact, a lot of the usage documentation lives elsewhere like in [the TypedTables docs](https://typedtables.juliadata.org/dev/man/group/). Contributions are welcome :)

Ah thanks. Another strategy I tried a while back was to first assume the join keys are distinct and then bail to a more general implementation when that's not the...

Yes - you are completely right. This *is* difficult. *And* the difficulty arises from the interaction of lots of little different aspects of the system, from the types, inference, standard...

Thanks @mbauman for the link. Yes, this is very interesting. I like the way this mixes up arrays, offset arrays, generalizations of dictionaries and Cartesian indices all in the one...

I have a swirling of thoughts in this space. Sorry, they are a bit disorganized. * Have we considered `A.axes.time` and `A.axes.channel`? We could make [this struct element](https://github.com/JuliaArrays/AxisArrays.jl/blob/master/src/core.jl#L170) a named...

Right. #81 looks like great work. I'm in favour of using `keys` and `axes` to control the behaviour of things like `similar`. For example, if `Base` always used `axes` instead...

I didn't follow - what does #58 kill? > (And yes, we're actually doing that in practice these days.) I can really see why you need something like AxisArrays to...