You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We have many algorithms (examples: copy!, vecdot, scale!, etc.) that take two array inputs and check whether they have the same number of elements, but do not otherwise care about their dimensionality or indices. Essentially, the two arrays are "auto-reshaped" to match.
I'm wondering whether we should change this, and in most cases require that their indices match. This is more feasible now that we have a non-copying reshape that works for arbitrary array types. The whole concern about efficiency discussed in http://julialang.org/blog/2016/03/arrays-iteration (under "ReshapedArrays") would be important; we'd probably want to have a paradigm for switching to the parent's iterator once the indices have been checked.
We have many algorithms (examples:
copy!
,vecdot
,scale!
, etc.) that take two array inputs and check whether they have the same number of elements, but do not otherwise care about their dimensionality or indices. Essentially, the two arrays are "auto-reshaped" to match.I'm wondering whether we should change this, and in most cases require that their
indices
match. This is more feasible now that we have a non-copyingreshape
that works for arbitrary array types. The whole concern about efficiency discussed in http://julialang.org/blog/2016/03/arrays-iteration (under "ReshapedArrays") would be important; we'd probably want to have a paradigm for switching to the parent's iterator once theindices
have been checked.Related: partial linear indexing, #14770.
The text was updated successfully, but these errors were encountered: