Implement XTensor support in core. #976
Conversation
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## master #976 +/- ##
==========================================
+ Coverage 86.48% 86.64% +0.15%
==========================================
Files 97 97
Lines 5951 5999 +48
==========================================
+ Hits 5147 5198 +51
+ Misses 804 801 -3 ☔ View full report in Codecov by Sentry. |
include/highfive/xtensor.hpp
Outdated
| val.resize(shape); | ||
| } |
There was a problem hiding this comment.
None of it works with non-trivial elements. I should check if they even allow anything non-trivial.
There was a problem hiding this comment.
We're now asserting the fact. I propose we revisit, if we ever need to support them.
|
@alkino The reason for not allowing non-trivial elements, is that this way we can use It doesn't work for anything that contains non-trivial elements. |
The commit removes the requirement for a constexpr rank. Instead
containers need a minimum and maximum constexpr rank.
The difficulties are:
* Empty arrays can't figure out their runtime rank. This is an issue
when deducing the dimension when writing to disk. The "solution" is
to deduce with lowest rank possible.
Adds support for `xt::xtensor`, `xt::xarray` and `xt::xview`, both row and column major. This works by wrapping the internal row-major with `xt::adapt`. Therefore, the `T` in `xt::xtensor<T, ...>` must be scalar (trivial).
Adds support for
xt::xtensor, both row and column major. We're missingxt::xarrayand views. Views fail because they claim to have runtime variable rank.