FailedChanges

Summary

  1. Add script to start S3 test container (details)
  2. Fix null exceptions in S3Layer classes (details)
Commit 3d531d22ce58ec729de9bf0f4f9fa9129952a561 by andy
Add script to start S3 test container
The file was addedscripts/runTestS3.sh
Commit c146f4ec957705c5ee49595f77c7986606e3ffd7 by andy
Fix null exceptions in S3Layer classes
...when performing operations against a bucket with no prefix, e.g.
"s3://example-bucket".
Fixed by preventing AmazonS3URI.getKey from returning nulls. This
_should_ be safe as a null path prefix is functionally equivalent to an
empty string in that they both represent
"no path prefix".
This change was made to reduce the chance that nulls escape downstream
into our various Layer[Writer|Reader|Copier|...] methods that don't have
a consistent way to handle a null prefix.
We also check for null in S3AttributeStore as a layered defense. More
aggressive checking in each of the downstream S3Layer classes would be
prohibitive ATM because the bucket/prefix information is copied in each.
Users should prefer use of the apply method over the class constructor
in order to take advantage of this safety check.
The file was modifieds3/src/main/scala/geotrellis/store/s3/S3AttributeStore.scala (diff)
The file was modifieddocs/CHANGELOG.rst (diff)
The file was modifieds3/src/main/java/geotrellis/store/s3/AmazonS3URI.java (diff)
The file was addeds3-spark/src/test/scala/geotrellis/spark/store/s3/S3NoPrefixSpec.scala