Antoine Grouazel

Yes I can think about a test. Could this test be a simple .tiff file opening or is it too high level?

I added a test in `test_backends.py` : `pytest -s ./test_backends.py::TestRasterio::test_rasterio_complex_dtype` But since there is a coming fix in `rasterio` I am wondering whether this PR is still needed: https://github.com/mapbox/rasterio/commit/7114fb7fb5e48146c4f04eb818daaa1fa632d817

Well @dcherian summed it up perfectly, it would be nice to have each single dataset collapsible.

I understood that the `a)` option won. Just to mention that the option c) is definitely the one that would be the best from my point of view. the "binding"...

I have to dig in the code to understood why we have this `BlockingActorProxy`. If it appears that it is no more needed I would be happy to remove it....

I will try to have a look at this issue next week.

issue opened in the reader repo : https://github.com/umr-lops/xarray-safe-rcm/issues/86

issue not yet fixed but at least PR #173 will help 1. the error raised to users is more clear 2. the comments in the code don't let think that...

@vincelhx I need clarifications, do you suggest modifications in https://github.com/umr-lops/xsar/blob/b0d5951f6125262eda252642c3dbbf085a415c0b/src/xsar/base_dataset.py#L711 ? Your suggestion is to use https://numpy.org/doc/stable/reference/generated/numpy.interp.html instead of https://docs.scipy.org/doc/scipy/reference/generated/scipy.interpolate.RectBivariateSpline.html ? Which raster products would use `interp` and which `RectBivariateSpline`...

I agree https://github.com/umr-lops/xsar/blob/b0d5951f6125262eda252642c3dbbf085a415c0b/src/xsar/base_dataset.py#L711 could be removed from `xsar` to be versioned in a separate repository. The new repository would be a dependency for `xsar` sphinx documentation compilation (especially `projections.ipynb`)