Review after Friday Jul 12th, have to meet with Pedro to ensure we extract all the necessary metadata.
The region_search tests pass DatasetRefs as dataIds - this is hot-patched in MockedButler but is something that we should probably fix before merging.
These changes to ButlerStandardizer improve its performance. Standardizes A0a, ~400 images, in ~94 seconds or so. Metadata is extracted via objects that do not load image, but just header metadata. This could additionally be cleaned up to save even more time per dataset reference, by constructing bounding box out of the corners of the image in summaryStats instead of calculating them via WCS.
These changes make the old ButlerStandardizer image collection unusable anymore, but given the costs to recreate those in this PR, I hope this is not a big problem. Minor changes to other functionality, more specific tests for raised errors and documentation.
Future change should probably be to just unroll the bounding box from objects into columns for ease of use.
Review after Friday Jul 12th, have to meet with Pedro to ensure we extract all the necessary metadata.
The region_search tests pass
DatasetRefs asdataIds- this is hot-patched inMockedButlerbut is something that we should probably fix before merging.These changes to
ButlerStandardizerimprove its performance. StandardizesA0a, ~400 images, in ~94 seconds or so. Metadata is extracted via objects that do not load image, but just header metadata. This could additionally be cleaned up to save even more time per dataset reference, by constructing bounding box out of the corners of the image insummaryStatsinstead of calculating them via WCS.These changes make the old
ButlerStandardizerimage collection unusable anymore, but given the costs to recreate those in this PR, I hope this is not a big problem. Minor changes to other functionality, more specific tests for raised errors and documentation.Future change should probably be to just unroll the bounding box from objects into columns for ease of use.