deprecate array-reducing isinteger #19925
Conversation
| @@ -2,8 +2,6 @@ | |||
|
|
|||
| ## Basic functions ## | |||
|
|
|||
| isinteger(x::AbstractArray) = all(isinteger,x) | |||
| isinteger{T<:Integer,n}(x::AbstractArray{T,n}) = true | |||
There was a problem hiding this comment.
With this change, we will lose this optimization. It seems to be blocked by the bounds check in all; adding @inbounds to all allows the loop to be optimized away.
There was a problem hiding this comment.
We might want to add
all{T<:Integer}(::typeof(isinteger), ::AbstractArray{T}) = true
all{T<:Real}(::typeof(isreal), ::AbstractArray{T}) = true
There was a problem hiding this comment.
As opposed or in addition to decorating all with @inbounds as you suggest above? Thanks!
There was a problem hiding this comment.
I think adding @inbounds to all is not safe unless we add a specialization for Array (where we know iteration never goes out of bounds).
There was a problem hiding this comment.
Specializations added. Thanks!
|
Also |
Re. |
|
There seems to be a good argument to keep
|
|
Added one commit deprecating vectorized |
Sacha0
changed the title
|
Should |
Per the docstring ( |
|
If |
|
It seems misleading to me to claim |
|
Saying "0 is an imaginary number" seems mildly crazy to me, but then again I'm not sure what this function is even for. There are no uses in Base. Are there example use cases that clarify which behavior would be desired? |
|
Also @Sacha0 you'll probably want to move the |
Sounds wise. Dropping the last commit and opening an issue re. |
|
Last commit dropped and issue #19947 opened. Are the vectorized |
Sacha0
changed the title
|
Upon reflection, I don't think we should deprecate
|
|
I separated the array-reducing |
Sacha0
changed the title
| isreal(x::AbstractArray) = all(isreal,x) | ||
| iszero(x::AbstractArray) = all(iszero,x) | ||
| isreal{T<:Real,n}(x::AbstractArray{T,n}) = true | ||
| all{T<:Integer}(::typeof(isinteger), ::AbstractArray{T}) = true | ||
| ctranspose(a::AbstractArray) = error("ctranspose not implemented for $(typeof(a)). Consider adding parentheses, e.g. A*(B*C') instead of A*B*C' to avoid explicit calculation of the transposed matrix.") | ||
| transpose(a::AbstractArray) = error("transpose not implemented for $(typeof(a)). Consider adding parentheses, e.g. A*(B*C.') instead of A*B*C' to avoid explicit calculation of the transposed matrix.") |
There was a problem hiding this comment.
bump for rebase. I think these two fallbacks are gone on master? or just moved, not positive
|
Thanks all! |
This pull request deprecates array-reducing
isinteger(A::AbstractArray)in favor ofall(isinteger, A)(edit: likewiseisreal). Discussion: #19674 (comment), #19598 (comment), and #19598 (comment). (Edit: Also fixesisimagfor real numbers that are zero.) Best!