Skip to content

Latest commit

 

History

History
76 lines (62 loc) · 6.51 KB

CHANGELOG.md

File metadata and controls

76 lines (62 loc) · 6.51 KB

Changelog

0.5.0 (2023-09-19)

  • Breaking change in specific scenarios: rescale=True now uses STAC metadata from raster:bands, instead of metadata set per GeoTIFF @RSchueder
  • Fix black stripes when mosaicing (introduced in 0.4.4) @gjoseph92 @lhcorralo
  • Update examples to point to newer Sentinel-2 endpoint on AWS @ljstrnadiii
  • Ensure jupyter-server-proxy is installed with pip install 'stackstac[viz]' @gjoseph92

0.4.4 (2023-06-21)

  • Resolve compatibility with NumPy >= 1.24.0 @gjoseph92
  • Fix TypeError when no items overlap with the given bounds @gjoseph92
  • Fix scale and offset application when only one is available @sharkinspatial
  • Change timer logging to DEBUG level @jorge-cervest

0.4.3 (2022-09-14)

  • Support sequences of Items @gjoseph92
  • Fix compatibility with pyproj>=3.4.0 @gjoseph92
  • stackstac has switched from Poetry to PDM as its package manager. This does not affect users, only developers.

0.4.2 (2022-07-06)

0.4.1 (2022-04-15)

  • Use pd.Index instead of deprecated pd.Float64Index @TomAugspurger
  • Better error when forgetting a custom nodata value to mosaic @gjoseph92

0.4.0 (2022-03-16)

  • Support specifying a chunk pattern for the time/band dimensions, allowing you to load multiple items in one spatial chunk (like stackstac.stack(..., chunksize=(-1, 1, 512, 512))). This can moderately to significantly decrease memory usage and improve performance when creating composites (like stack(..., chunksize=(-1, 1, "128MiB", "128MiB")).median("time")). See #116 (comment) for details.
  • stackstac.mosaic generates a more efficient dask graph, with hopefully lower memory usage @gjoseph92
  • Clearer errors when versions of pystac or satstac are incompatible @gjoseph92
  • Support newer versions of xarray using CalVer @scottyhq

0.3.1 (2022-01-20)

0.3.0 (2022-01-20)

  • Breaking change: fill_value=None is no longer supported. You must always specify a fill_value (default is still NaN); it can no longer be inferred from the GeoTIFF files. @gjoseph92
  • Respect fill_value for array chunks that don't overlap with any Asset, instead of always using NaN @gjoseph92 @TomAugspurger
  • Fix bugs with stackstac.show when the path to your notebook file had the word notebook, lab, or voila in it @robintw
  • Support 2022 version of Dask @gjoseph92
  • Relax NumPy requirement, supporting any NumPy version supported by Dask @gjoseph92 @scottyhq
  • Require minimum Pillow version of 9.0, due to vulnerabilities reported in older versions (in features unused by stackstac)

0.2.2 (2021-12-03)

0.2.1 (2021-05-07)

Support xarray 0.18 and beyond, as well as looser version requirements for some other dependencies.

0.2.0 (2021-05-05)

Call stackstac.show to render DataArrays in interactive ipyleaflet maps in your notebook! See this example for more.

  • mosaic function
  • Exposed some assorted handy geospatial ops
  • More robustly get band metadata (if one Item is missing some asset metadata, just ignore it instead of dropping the whole thing)
  • Fixed occasional ValueError: conflicting sizes for dimension
  • Resolved issue where timestamps became integers
  • Support pystac-client ItemCollections
  • Minimum Python version is now (accurately) 3.8. Stackstac would have failed upon import on 3.7 before.

0.1.1 (2021-04-16)

  • Passing a gdal_env now works when using a distributed cluster (before, you got a pickle error when calling .compute())
  • Many typos fixed, thanks @kylebarron and @RichardScottOZ!

0.1.0 (2021-03-10)

Initial release 🎉