Releases: jimy-byerley/arrex
Releases · jimy-byerley/arrex
v0.5.2
v0.5.1
- fixed wrong struct format for
glm.TmatN
- fixed processor's struct padding always set on 8 bytes, it now checks the actual word size
v0.5.0
- added missing support for index and slice deletition in
typedlist
v0.4.2
- fixed segfault in typedlist slicing when given stop < start
v0.4.0
improved support for pickle
- now, arrays can be pickled as soon as their dtype can be pickled
- dtypes that can be pickled are auto-declarant dtypes (like structs and python classes) and compiled dtypes that can be imported automatically
v0.3.2
v0.3.1
- bug fixes in unsigned comparisons in some methods
- fixed missing cython file in the package, causing the compilation to fail if cython were present
v0.3
This update brings many fixes:
- fixed memleak in item access (
typedlist.__getitem__
) - fixed memleaks in
typedlist.index
,typedlist.reverse
- fixed concatenation with buffers with a size not multiple of
dtype.size
that could lead to ambiguous behaviors
v0.2
This release is much more usable than the previous one. It brings the following:
- a complete rework of the concept of dtype
- a dtype system allowing pretty much anything to be a dtype, declared through a ddtype (see documentation coming)
- a native way to use python ints/floats in
typedlist
- stronger tests
- few bug fixes
v0.1
That first version provides an overview of the module functions:
typedarray a dynamically sized array
- ref-counted internal buffer
- flexible dtype, not known by advance
- copyless views and slices