v3.2.0

Remove trailing spaces in docs.

v3.2.0

What's Changed

• document index_chunks in example by @lmiq in #62
• size for RoundRobin() by @lmiq in #63

Full Changelog: v3.1.2...v3.2.0

v3.1.2

ChunkSplitters v3.1.2

Diff since v3.1.1

https://github.com/JuliaFolds2/ChunkSplitters.jl/blob/main/CHANGELOG.md

Merged pull requests:

• out of bounds error on empty iterator indexing (#59) (@lmiq)
• Allow minsize to exceed the collection length (#60) (@MasonProtter)

Closed issues:

• What should happen if length of collection is less than n? (#58)

v3.1.1

ChunkSplitters v3.1.1

Diff since v3.1.0

• throw error if minsize > length(collection).

Merged pull requests:

• Do not CI run 1.6 on MacOS-latest (#54) (@lmiq)
• CI docs on lts and update CI versions (#55) (@lmiq)
• throw error if minsize > length(collection) (#57) (@lmiq)

Closed issues:

• Speculative: should we return chunks of the object instead of chunks of indices? (#25)
• Thread-safe, Channel-like, iteration of chunks (#49)
• Empty iterator if size(collection) < minsize (#56)

v3.1.0

ChunkSplitters v3.1.0

Diff since v3.0.0

• Recover compatibility with Julia 1.6 to 1.9. On Julia 1.6 an allocation test fails. At 1.9 all tests pass.

Merged pull requests:

• recover support for older Julia versions (#52) (@lmiq)

v3.0.0

ChunkSplitters v3.0.0

Diff since v2.6.0

This release is a major overhaul of the user interface.

Changes:

• The new lower compat bound for Julia is 1.10. This implies that support for the old LTS 1.6 has been dropped.
• The old main API function chunks is deprecated, and the new chunks function behaves differently.
• The new chunks has been introduced that returns an iterator that provides chunks of elements rather than chunks of indices of a given input collection. To avoid copies, it is based on view.
• A new function index_chunks is introduced that returns the indices of the chunks.
• For performance reasons, the split keyword option now requires a Split instead of a Symbol. Concretely, :batch should be replaced by Consecutive() and :scatter should be replaced by RoundRobin().
• The keyword argument minchunksize has been renamed to minsize.
• getchunk isn't public API anymore (and has, internally, been renamed to getchunkindices). If you really need a replacement, consider using index_chunks(...)[i] instead.
• The old legacy API that relied on positional rather than keyword arguments has been dropped.

Merged pull requests:

• Changes for 3.0 (#48) (@carstenbauer)
• Simplify iterate method to remove allocation on 1.10. (#50) (@fredrikekre)
• Simplify enumerate iterate (#51) (@lmiq)

Closed issues:

• 3.0 (#47)

v2.6.0

ChunkSplitters v2.6.0

Diff since v2.5.0

New features:

• add minchunksize option.
• add BatchSplit and ScatterSplit types as (type-stable) default alternatives to symbols :batch and :scatter. Types are preferred and should be used to guarantee that the generation of the chunks do not allocate any intermediate.

Merged pull requests:

• minchunksize + split::SplitStrategy (new default) (#46) (@carstenbauer)

v2.5.0

ChunkSplitters v2.5.0

Diff since v2.4.5

• Support eachindex(enumerate(chunks(...)))

Merged pull requests:

• implement eachindex(enumerate(...)) (#42) (@lmiq)
• Drop Compat dependency + Update CI (#44) (@carstenbauer)

Closed issues:

• Support eachindex for ChunkSplitters.Enumerate (#41)
• Drop support for Julia 1.6 (#45)

v2.4.5

ChunkSplitters v2.4.5

Diff since v2.4.4

Merged pull requests:

• New version: v2.4.5 (#39) (@lmiq)

v2.4.4

ChunkSplitters v2.4.4

Diff since v2.4.3

• Fixed #36.

Merged pull requests:

• Empty range instead of nothing when getchunk returns (#38) (@lmiq)