Enable reshape_generic for slices + other things
#1180
Conversation
|
CI failure seems to be due to a spurious failure of an eigen proptest. |
|
Now that #1178 was merged, I will endeavor to update this PR soon. @sebcrozet: any thoughts on whether you would want to merge all 3 parts of this PR, or if I should leave some out (either for a separate PR or not at all, if the changes are not desirable). |
|
@Andlon Thank you for this PR! Keeping the 3 parts of this PR is fine. I agree with deprecating |
sebcrozet
force-pushed
the
reshape_slices_plus_more
branch
from
3931728
to
72813ea
Compare
This allows us to simply write U4 in place of U4::name() or Const::<4>, like we used to be able to before const generics.
Provide a type alias to avoid breaking code. Make Dyn a tuple struct so that we can use the succinct syntax Dyn(n) instead of Dyn::new(n).
In the process of implementing ReshapbleStorage for SliceStorage(Mut), I discovered that there appears to be no tests for the existing reshape_generic functionality on owned matrices.
sebcrozet
force-pushed
the
reshape_slices_plus_more
branch
from
dc59ef7
to
711ac67
Compare
|
@Andlon I took the liberty to update the PR and deprecate |
|
@sebcrozet: that's amazing, thank you! Pretty swamped with work and life at the moment, so haven't been able to dedicate any time to this. Great to see it merged and in the new release :D |
This PR consists of 3 commits, which can be split out as different PRs as necessary:
U0, U1, ...constants alongside theU0, U1, ...types. This lets us writeU4instead ofU4::name()orConst::<4>when we need const dimensions. It used to be like this before we got const generics, but it got lost in the transition. I don't think there's any implied breakage from this.DynamictoDynand makeDyna tuple struct.Dyn(3)instead ofDynamic::new(3), which is very verbose.Matrix<T, _, Dynamic, VecStorage<...>>more concise, which might make them easier to read.Dynamic = Dynis also provided.Dynamic.ReshapableStoragefor matrix slices (only for unit strides at the moment), which was missing.slice.reshape_generic(...)on slices as well, something I've needed many times.reshape_genericfunctionality, so I added some.This PR can either be merged before #1178, in which case I'd update #1178, or it can be merged afterwards, in which case I'll update this PR with the new view terminology.