The rabbit hole that is the data macro about dataarrays.jl HOT 6 OPEN

Damn. In the discussion I started on julia-stats a while ago [1], @StefanKarpinski said that he thought [1,NA,2] should be possible at some point, using later-stage functions. Since this means @data wouldn't be needed anymore, I'd say just ignore this for now, and let people use the data() function.

1: https://groups.google.com/d/topic/julia-stats/1CYmgu1Lye0/discussion

from dataarrays.jl.

I hope staged functions are sufficient. I am going to ignore this for now since I don't see a way forward without new tools in Base.

from dataarrays.jl.

The right way to do this is to implement datahcat, datavcat, and datahvcat and make the data and pdata macros call these. I have this working, but 1) if we care about accurate type inference, it's a decent amount of code copy/pasted from Base with a few lines changed and 2) it's not currently working because of a bug in setindex!.

from dataarrays.jl.

Oh great. I had naively assumed the *cat functions were builtins.

So much of our codebase is basically copy/pasted from other places that I've got no opposition to doing it again.

What's the bug in setindex!?

from dataarrays.jl.

The bug is that setting a portion of a DataArray to be another DataArray fails if the latter has NAs. #62 has an easy fix that would unfortunately create a bunch of ambiguity warnings if applied to master, but it isn't the complete fix; we should also handle setting a portion of a DataArray to be an arbitrary AbstractDataArray.

from dataarrays.jl.

(And after that change, my tests are still failing, now because of JuliaLang/julia#5448.)

from dataarrays.jl.