Paul Haesler

I suspect it's related to over-writing the product band index in the product/layer configuration object (from #1041): ``` req.product.band_idx = FakeBandIndex() # very hacky... ``` These objects are created on...

I think this was introduced in https://github.com/opendatacube/datacube-core/pull/1912 - pretty sure it's the thing Peter was confused by and asked me to clarify but you guys merged it before I had...

> on a less explicit check e.g. the weird hack that used to be there before #1912, or wrap the import in a Deprecation warning suppression, as used in tests:...

Can we do the check in reverse? None: cool, do nothing. isinstance(odc.geo.geobox.GeoBox): cool, do nothing. Anything else: assume legacy geobox and convert?

Thanks Christoph. As it happens we've had some very similar conversations around WMS/WCS interoperation internally at DEA recently. The problem is that WMS and WCS are very different protocols with...

Nice - yes that's one (slightly hacky) way of doing it! I would like to handle this more generically at some point by treating the index functions of all defined...

That whole schema has disappeared in 1.9 because we no longer support the ingestion workflow.

1.9 branch is fully SQLAlchemy2 compatible. I think the resolution of this issue is retiring 1.8

Shapely knows nothing about CRS's and doesn't handle coordinate transforms, so it's about what PyProj can support rather than Shapely. However, PyProj DOES support 3D coordinate transforms, so this should...

No idea what is happening here, but looks like the new GDAL version is breaking nodata handling somehow. Hold off merging until we at least understand what's going on. (I...