Changelog

Unreleased

Import titiler.xarray (from titiler repo) (#72)
Rename the package to titiler.multidim (#72) breaking change
Drop support for kerchunk reference files (#72) breaking change
Drop support for experimental multiscale zarr group zoom level functionality (#72) breaking change
Remove default WebMercatorQuad tile matrix set in /tiles, /tilesjson.json, /map and /WMTSCapabilities.xml endpoints (with upgrade to titiler.core>=0.19) breaking change
Use uv for managing dependencies #74
Slim down the Lambda asset package size #74
- run strip on compiled C/C++ extensions (except numpy.libs)

v0.2.0

Improved pyramid support through group parameter

Add support for a group parameter in /histogram route.
Catch zarr.errors.GroupNotFoundError and raise 422 in the tiles route. When the multiscale parameter is true but the zoom level doesn't exist as a group in the zarr hierarchy, this error is raised.

Add metadata caching via redis cache and AWS elasticache

Added metadata caching via redis cache and AWS elasticache.
Use fakeredis for cache in tests.
Remove starlette-cramjam CompressionMiddleware.
Address more cases of protocol/engine combinations in reader.py#get_filesystem.
Moved cftime and pandas requirements from Dockerfile to pyproject.toml.

v0.1.1

Support for NetCDF and making consolidated metadata optional. See #39.

Performance results between prod (v0.1.0) and dev (unreleased).

Performance for supported datasets is about the same.
Unsupported datasets in v0.1.0 (netcdf and unconsolidated metadata) reported 100% errors in prod and 0% in dev (expected).
- NetCDF Dataset: pr_day_ACCESS-CM2_historical_r1i1p1f1_gn_1950.nc
- Unconsolidated metadata dataset: prod-giovanni-cache-GPM_3IMERGHH_06_precipitationCal

v0.1.0 (2023-10-11)

Initial release of the project.