Releases: JuliaGPU/Adapt.jl
v4.1.1
v4.1.0
v4.0.4
Adapt v4.0.4
Merged pull requests:
- Specialize adapt_structure for small tuples (#81) (@charleskawczynski)
Closed issues:
- Inference failure in
adapt
(#79)
v4.0.3
Adapt v4.0.3
Merged pull requests:
v4.0.2
Adapt v4.0.2
Merged pull requests:
- Bump codecov/codecov-action from 3 to 4 (#76) (@dependabot[bot])
- Bump actions/cache from 3 to 4 (#77) (@dependabot[bot])
- Use recursion to fix inference failure (#78) (@charleskawczynski)
Closed issues:
- Inference failure in
adapt
(#79)
v4.0.0
Adapt v4.0.0
Breaking changes: Adapt.ndims
, Adapt.eltype
and Adapt.parent
(type-versions of Base functions) have been removed. Instead, Base.ndims
and Base.eltype
should be used, which works fine on typed. To replace Adapt.parent
, use the new and exported parent_type
or unwrap_type
functions (the former only returns the direct parent, while the latter returns the innermost type). Note that where Adapt.parent
used to return a raw typename, parent_type
and unwrap_type
return type, i.e., with type variables retained.
Note that these changes likely don't affect you, as there are no public users of these interfaces.
Merged pull requests:
Closed issues:
- Rename
Adapt.ndims
for clarity (#72)
v3.7.2
v3.7.1
Adapt v3.7.1
Merged pull requests:
v3.7.0
Adapt v3.7.0
Merged pull requests:
- Implement missing LinearAlgebra wrappers and add support for uplo parameter (#51) (@danielwe)
- enable dependabot for GitHub actions (#65) (@ranocha)
- Bump actions/checkout from 2 to 3 (#66) (@dependabot[bot])
- Bump actions/cache from 1 to 3 (#67) (@dependabot[bot])
- Bump codecov/codecov-action from 1 to 3 (#68) (@dependabot[bot])
- Bump actions/checkout from 3 to 4 (#69) (@dependabot[bot])