[Diff since v0.14.3](v0.14.3...v0.14.4)

**Merged pull requests:**
- Throw informative error for unsupported vtype (#255) (@Sbozzolo)

**Closed issues:**
- More informative changelogs for breaking releases (#242)
- is there a variant of `load!` that accumulates? (#252)
- Issues with OpenDAP (#253)
- Run Aqua tests (#254)

[Diff since v0.14.2](v0.14.2...v0.14.3)

* allow CartesianIndex and CartesianIndices with load! (issue #250, thanks to @haakon-e)

[Diff since v0.14.1](v0.14.1...v0.14.2)

[Diff since v0.14.0](v0.14.0...v0.14.1)

[Diff since v0.13.2](v0.13.2...v0.14.0)

* `chunksize` is now a tuple of integers (was a vector of integers). On input (of `defVar` and `chunking`), a vector of integers is still accepted for compatibility.
* `idimensions` key word argument is dropped from write. Use `SubDatasets` instead to write a subset of a netcdf file.
* Many functions and types not specific to NetCDF files (multi-files, views, and load by value) have been moved to [CommonDataModel.jl](https://github.com/JuliaGeo/CommonDataModel.jl)

[Diff since v0.13.1](v0.13.1...v0.13.2)

**Merged pull requests:**
- NetCDF3 files (64-bit offset NetCDF,...) and unlimited dimensions (issue #231) (#232) (@Alexander-Barth)
- Minor docs improvements (#236) (@glwagner)
- Corrected the SST image output (#239) (@keduba)

**Closed issues:**
- 64-bit offset NetCDF and unlimited dimensions (#231)
- .var[:] always returns a vector (#233)
- check presence of attribute and variables (#235)
- String-valued dimension incorrectly loaded as matrix of characters (#237)

[Diff since v0.13.0](v0.13.0...v0.13.1)

[Diff since v0.12.17](v0.12.17...v0.13.0)

NCDatasets uses now the DiskArray package which required some API changes.
In general, the array API of NCDatasets is now more similar to base Julia in particular:

* `ncvar[range_indices] = scalar` should now be  `ncvar[range_indices] .= scalar`
* `ncvar2D[:]` flattens the data in the 2D NetCDF variable `ncvar2D`. To read the full array one need to use `ncvar2D[:,:]` or `Array(ncvar2D)` (similarly for 3D, 4D... arrays).
* Accessing an array out of bounds, new returns a `DimensionMismatch` exception (previously a `NCDatasets.NetCDFError` exception was returned)
* To grow a NetCDF variable with unlimited dimension, the corresponding index of left-hand side of the assignment cannot be a colon, but should be a range. For example if `ncvar` is a NetCDF variable where the 2nd dimension is unlimited, `ncvar[:,:] = zeros(2,3)` should now be replaced by `ncvar[:,1:3] = zeros(2,3)`

**Merged pull requests:**
- DiskArrays for `Variable`'s (#205) (@tcarion)
- Fix checksum docstring (#213) (@navidcy)
- Correction of typos (#216) (@keduba)
- CompatHelper: add new compat entry for "DiskArrays" at version "0.3" (#229) (@github-actions[bot])

**Closed issues:**
- NCDatasets.jl v0.11+ does not work on PowerPC  (#116)
- For compressed netcdf, extremely low speed in cdo operation (#206)
- LoadError: UndefVarError: `NetCDFError` not defined (#207)
- [Suggestion] easier creation of time axis (#208)
- [enhancement] support empty array `var[ [] ]` (#209)
- Using NCDatasets.jl in R (#210)
- No warning message when passing wrong keywords to `defVar` (#212)
- @select not work for 4d array (#214)
- NCDatasets fails due to build issue on Julia 1.10 (#215)
- Significant performance gap between NetCDF.jl and NCDatasets.jl (#218)
- Write data fails when the number less than typemax(Datatype) (#226)
- Return type `Matrix{Union{Missing, Float32}}`  (#227)

[Diff since v0.12.16](v0.12.16...v0.12.17)

[Diff since v0.12.15](v0.12.15...v0.12.16)