lilithhafner/ptrarrays.jl

Back to the stone age!

ptrarrays.jl's Issues

DenseArray - support for strides

I was wondering if it would be possible to bump PtrArray from a subtype of AbstractArray to a subtype of DenseArray. At first glance, as the memory layout is in fact dense and strided, it might make sense to make that explicit.

For me specifically, I would like to play around with these array types in the context of TensorOperations.jl and Strided.jl. There, contractions of multiple tensors often involves temporary array objects, and TensorOperations itself already inserts alloc and free calls within the generated code. It should thus be quite easy to plug PtrArrays in there, but the current implementation uses Strided.jl, which works on DenseArrays.

If this makes any sense, I'd be happy to drop a PR.

Should throw when length overflows

julia> using MallocArrays

julia> x = malloc(Int, 10000000, 10000000, 10000000, 100000000);

julia> typeof(x)
MallocArray{Int64, 4}

julia> size(x)
(10000000, 10000000, 10000000, 100000000)

julia> length(x)
7886392056514347008

julia> x[1]

[301693] signal 11 (1): Segmentation fault
in expression starting at REPL[7]:1
Segmentation fault (core dumped)

TagBot trigger issue

This issue is used to trigger TagBot; feel free to unsubscribe.

If you haven't already, you should update your TagBot.yml to include issue comment triggers.
Please see this post on Discourse for instructions and more details.

If you'd like for me to do this for you, comment TagBot fix on this issue.
I'll open a PR within a few hours, please be patient!

Change license to e.g. MIT?

Since AliasTables depends on PtrArrays, the license of PtrArrays is problematic for downstream packages such as Distributions or StatsBase that want to avoid (or at least minimize) GPL dependencies. See e.g. JuliaStats/Distributions.jl#1864. Based on JuliaStats/Distributions.jl#1864 (comment), I assume it might be possible to relicense PtrArrays under the MIT license?