0.10.0-RC3 is released

[Rob Emanuele]

GeoTrellis 0.10.0-RC3 is released and on sonatype. As always, it'll take a little bit to propagate so it might be a little while before it's available on maven central.

There have been a number of fixes and final adjustments between RC2 and RC3. The only API breaking changes that have happened are as follows:

- "geotrellis.spark.ingest.Pyramid" has been moved to "geotrellis.spark.pyramid.Pyramid"

- The rasterizer API has been changed slightly:

   - non-useful functionality removed

   - useful overloads added

   - renaming of "rasterize" to "rasterizeWithValue" in cases where you are passing in a burn value

   - reordering of parameters to have the options at the end.

   - Rasterize.Options was renamed to Rasterizer.Options

- Component types moved from "geotrellis.spark" to "geotrellis.util"

- Changed "renderGeoTiff", "renderPng" and "renderJpg" methods on RDDs to return GeoTiff, Png and Jpg elements instead of Array[Byte]

Other changes in this release:

 - Fixed an issue with the fallback color not being properly set in color map PNGs

 - Fixed RGB and greyscale + alpha PNG encoding

 - Fixed issue with S3 template URL parsing when used with the "saveToS3" method

 - Implemented NTV1 grid file parsing in geotrellis.proj

 - Fixed floating point errors on no-op reprojection of RDDs

 - Implemented functionality for adding splits to an Accumulo table based on KeyBounds,

   which is done automatically on tables created by writing a layer. This will improve

   Accumulo ingest speeds.

 - Fixed a bug in union and intersect operations on CellType

 - Fixed bug in how ColorRamps were coloring values at class border stop points.

 - Fixed bug in merging RDDs of tiles with BitCellType

 - Fixed a bug in reading GeoTiffs with INTERLEAVE=BAND and PREDICTOR=2 or 3

 - Made the Tile.convert method non-lazy, removed LazyConvertedArrayTile as it was causing many issues around proper handling of cell type conversions.

 - Removed cloud removal code, as it wasn't quite at the generality and robustness of the library. You can find the code here https://gist.github.com/lossyrob/23560c1d6953943b2847

 - Moved the Component architecture away from monocle lenses, and created GetComponent and SetComponent, so that context bounds can request read-only components.

 - Don't use App anymore in ETL or examples, it can cause evil initialization bugs when used with Spark and we recommend staying away from it.

 - Removed unimplemented ETL options.

 - Made small change to allow Java serialization to be used the library, except for the case of reading and writing to Accumulo, which requires Kryo.

 - Allow PutObjectRequest modification when writing to S3, which allows users to set the ACL to a non-default (e.g. for public access)

 - Fixed bug in partitionCount option in ETL for S3 input

 - Set the ETL to keep partition count through tiling operation, so that if source tiles are manually repartitioned the partition count stays constant throughout the tiling process.

 - Removed unnecessary clamping in gridToMap functions on RasterExtent

 - Added "split" function, which allows splitting tiles inside an RDD into smaller chunks before processing. This is useful when dealing with large tiles.

 - Removed "isValid" check and potential polygon modification in Polygon object apply.

 - Fixed a bug in ETL "per-tile" reprojection.

Thanks to everyone who has contributed, and for everyone's patience as we put the final touches on 0.10.

Best,

Rob