Pietro Vertechi

> Ah yeah so you mean it would make sense to have `ShiftedArrays.diff` return a lazy array instead, with arguments to choose whether the first value should be missing or...

Thanks for the nice words! Nested structures are supported but unfortunately undocumented (I'm keeping the issue open to remember to add docs)... When calling `StructArray` you can specify with `unwrap`...

Yes, I'm closing #126 because nested structs are not broken and the approach suggested there work. Just ping me when you PR is ready for review (and of course if...

If storage is already column-based, there should be no allocations at all. In your case, I think the easiest is to define a custom helper function to construct things recursively,...

Performance improvements are always welcome, so I guess it'd be good to see where the timing difference comes from, and see if the same trick can be used here. These...

The constructor performance may have been fixed by #162 (just in time :) ). I guess it'd be useful to see what happens when you use a StructArray of tuples,...

> How is this related to ConstructionBase or [JuliaArrays/StructArrays.jl#241](https://github.com/JuliaArrays/StructArrays.jl/pull/241)? I don't see any type computations in ConstructionBase now, only construction/modification of instances. That PR also doesn't touch StructArrays eltype computations...

I wanted to ask if there were any developments on this issue. Would it be acceptable to simply assume (if the user does not provide a range) that the data...

Just to pitch in (as this was brought up in https://github.com/piever/StructArrays.jl/pull/84), I think that changing the eltype should be done not by adding a custom `eltype` method but by giving...

Somehow I'm not sure we should add some specific macros here: if they didn't make it to Remark.js, then it's just confusing to have them here. OTOH it could be...