baumgold opened a new pull request, #405:
URL: https://github.com/apache/arrow-julia/pull/405
Currently when there's a table containing a column of type variable-size
list, whenever the user accesses an entry in the list a new Vector is
instantiated causing allocations and performance degradation. This PR proposes
to prevent instantiating the Vector and instead return to the user a zero-copy
view of the data (SubArray) that still abides by the AbstractVector interface.
For example, instead of getting a newly-allocated `Vector{Int64}`:
```julia
julia> using Arrow, Tables
julia> x = (a=[1,2,3], b=[[1,1,1], [2,2], [3,3,3,3]]);
julia> table = Arrow.Table(Arrow.tobuffer(x))
Arrow.Table with 3 rows, 2 columns, and schema:
:a Int64
:b Vector{Int64} (alias for Array{Int64, 1})
julia> p = iterate(Tables.rows(table))
(Tables.ColumnsRow{Tables.CopiedColumns{Arrow.Table}}: (a = 1, b = [1, 1,
1]), 2)
julia> p[1].b
3-element Vector{Int64}:
1
1
1
```
we'll instead get a `SubArray`:
```julia
julia> using Arrow, Tables
julia> x = (a=[1,2,3], b=[[1,1,1], [2,2], [3,3,3,3]]);
julia> table = Arrow.Table(Arrow.tobuffer(x))
Arrow.Table with 3 rows, 2 columns, and schema:
:a Int64
:b Vector{Int64} (alias for Array{Int64, 1})
julia> p = iterate(Tables.rows(table))
(Tables.ColumnsRow{Tables.CopiedColumns{Arrow.Table}}: (a = 1, b = [1, 1,
1]), 2)
julia> p[1].b
3-element view(::Arrow.Primitive{Int64, Vector{Int64}}, 1:3) with eltype
Int64:
1
1
1
```
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
