Olivier Gauthé

Indeed we would have more freedom with `abstract type AbstractGradedUnitRange{T,CS} `. Just to be clear, that would be a significant change in the implementation. It seems to me that implies...

Here is a first draft passing `test_basics.jl`. Concerning `dual`, if I remember correctly, the rational behind `UnitRangeDual` was precisely to have `GradedUnitRange = BlockedUnitRange{LabelledInteger}`, at a time when `BlockArrays` did...

`GradedAxes` and `SymmetrySectors ` tests are passing. I am now working on issues with `BlockSparseArrays`. I spotted several cases where the labels are dropped: ```julia d1 = gradedrange([U1(0) => 2,...

Indeed the problems come from missing overload for `BlockArrays.combine_blockaxes`. Fixing it in a generic way should solve many related issues in ITensor/BlockSparseArrays.jl#2 .

The other problems where due to the change `blockedunitrange_getindices/gradedunitrange_getindices`. Maybe I should avoid defining `gradedunitrange_getindices` and just specialize `blockedunitrange_getindices` for `AbstractGradedUnitRange`.

I fixed all the tests for `GradedOneTo` and `GradedUnitRange`. I only define `blockedunitrange_getindex`. I am not very satisfied with the amount of code duplication for `blockedunitrange_getindex` but most versions are...

Thanks for the review. I already started working on a slicing `GradedUnitRangeDual`, but this is non trivial. I am pretty happy that many issues appearing in ITensor/BlockSparseArrays.jl#2 actually come from...

All tests are passing locally. The failures come from unchanged `UnallocatedArrays`, I suppose they are not related to this PR.

> @ogauthe looks like the test failures are fixed now, is this ready to merge once the rest of the tests pass? yes it is

Thank you for the fix. > btw, there's no actual computation in your `slice_reshape_transpose` code... even if I were to succeed implementing this without copy, I don't see where pythran...