dsharlet/array

Issues

• `array` and `array_ref` `operator==` and `operator!=` don't short circuit

  #4 opened 5 years ago by dsharlet
  7

• Annotate types with ABSL attributes?

  #107 opened 5 months ago by jiawen
  1

• Automatic strides for a final dim of extent 1 is 1 instead of product of previous extents in the default and dense_array case

  #101 opened 9 months ago by jiawen
  0

• whole-matrix einsum is slower than hand-written code

  #37 opened a year ago by dsharlet
  2

• 3x1x3 interleaved layout returns y().stride() of 1 instead of 9

  #91 opened 2 years ago by scribbleink
  4

• Consider adding "flat iterators" for arrays

  #22 opened 5 years ago by dsharlet
  5

• errors.cpp test is failing silently on gcc

  #92 opened 2 years ago by dsharlet
  0

• shape allows construction from incompatible shapes (std::is_constructible and std::is_assignable are broken for std::tuple on various STLs)

  #20 opened 5 years ago by dsharlet
  4

• array's move constructor default constructs an allocator before using the other allocator

  #47 opened 2 years ago by dsharlet
  0

• const_cast for array_ref<const T>

  #85 opened 2 years ago by jiawen
  8

• Cropping via ranges results in an array with the same indices as the old array

  #24 opened 5 years ago by dsharlet
  1

• Need a version of generate() that takes indices

  #57 opened 2 years ago by jiawen
  3

• Alias for a fixed size array allocated on the stack

  #65 opened 2 years ago by fbleibel-g
  6

• call of overloaded ‘for_each_value_in_order<(nda::shape<>::rank() - 1)>(nda::shape<>::index_type, void (&)(const int&, int&), std::pair<const int*, std::tuple<> >&, std::pair<int*, std::tuple<> >&)’ is ambiguous

  #67 opened 2 years ago by Mr-doooog
  4

• reinterpret_shape doesn't resolve bare shape

  #72 opened 2 years ago by linht122
  3

• `make` broken on Apple Silicon Macs with Xcode

  #70 opened 3 years ago by jiawen
  1

• runtime assert failures regarding incompatible dimension parameters are hard to debug

  #26 opened 3 years ago by dsharlet
  7

• Numpy -> nda::array cheat sheet?

  #63 opened 3 years ago by mgraczyk
  0

• Write a FAQ with examples of how to do common tasks

  #56 opened 3 years ago by jiawen
  1

• Reshape an array in runtime

  #50 opened 3 years ago by votientu
  1

• Template argument order issue in make_* functions that return an array

  #53 opened 3 years ago by dsharlet
  0

• Host doxygen generated docs somewhere

  #51 opened 4 years ago by dsharlet
  1

• Remove copies from Einstein expression construction

  #42 opened 5 years ago by dsharlet
  1

• Bounds issues should be reported consistently

  #10 opened 4 years ago by dsharlet
  0

• auto_allocator could fall back to a base allocator when the allocation is too big

  #46 opened 4 years ago by dsharlet
  0

• clang with -O3 generates significantly worse code than with -O2

  #36 opened 5 years ago by dsharlet
  1

• Comparison with mdspan/mdarray

  #45 opened 5 years ago by xelatihy
  1

• Consider removing make_dense* functions

  #43 opened 5 years ago by dsharlet
  0

• make_compact should give dimensions constexpr strides where possible

  #31 opened 5 years ago by dsharlet
  0

• Add CUDA support

  #30 opened 5 years ago by dsharlet
  1

• Implicit conversion to scalar value for scalar shaped arrays?

  #32 opened 5 years ago by dsharlet
  0

• Consider free function to access dims instead of member functions

  #34 opened 5 years ago by dsharlet
  0

• dim allows copy construction with compile-time incompatible strides

  #21 opened 5 years ago by dsharlet
  0

• Vcpkg (and cmake) support.

  #29 opened 5 years ago by acdemiralp
  1

• Find a good syntax for expressing ranges

  #25 opened 5 years ago by dsharlet
  3

• Better differentiate between ranges and shapes

  #23 opened 5 years ago by dsharlet
  0

• Shapes should only get automatically defined strides when they are used to construct arrays

  #17 opened 5 years ago by dsharlet
  0

• Automatic strides are broken if one of the strides is UNK

  #9 opened 5 years ago by dsharlet
  0

• stack_allocator should not have fake copy constructors/assignment operators

  #7 opened 5 years ago by dsharlet
  0

• It should be possible to be completely oblivious to mins

  #18 opened 5 years ago by dsharlet
  0

• `array::set_shape` is unsafe

  #19 opened 5 years ago by dsharlet
  0

• reshape operations don't check bounds

  #11 opened 5 years ago by dsharlet
  0

• Should move/make_*_move use rvalue references for the source?

  #8 opened 5 years ago by dsharlet
  0

• Error messages can be unwieldy

  #15 opened 5 years ago by dsharlet
  1

• Rename stack_allocator

  #6 opened 5 years ago by dsharlet
  0

• Slicing does not reduce rank

  #14 opened 5 years ago by dsharlet
  0

• Shape constructors are ambiguous for 1D shapes

  #16 opened 5 years ago by dsharlet
  0

• at() doesn't throw if the index is out of bounds

  #13 opened 5 years ago by dsharlet
  0

• GCC 8.3.0 is almost 2x slower than clang 10.0.0 in performance tests

  #12 opened 5 years ago by dsharlet
  1

• `array::swap` does not respect `std::allocator_traits<Alloc>::propagate_on_container_swap::value`

  #5 opened 5 years ago by dsharlet
  0