Martin Raspaud

Welcome to pytroll and thank you for your contribution @dpfinch ! I see that the tests are failing, since we need to support xarray.DataArray and dask.array.Array structures too...

@rouault thanks you very much for taking a look at this. I can confirm that using `ellps=sphere` solves the problem. Maybe it would be a good thing to add a...

Does anyone have a workaround for this? We would just like to run missing benchmarks when the server is idle...

Just replying to this thread, way later, sorry. I just wanted to mention that pyresample is in LGPL license, if that makes a difference. Regarding the developer week, it would...

I'm working supporting generic dask resampling as we speak: https://github.com/pytroll/pyresample/pull/341

TBH, I can't see any difference. My guess is that somehow the gil was already released... before the change (so not releasing the gil?)  after the change (releasing the...

I did recompile! Checking now with other settings

Tried with 3 composites, a bigger destination area, the setting was already to `dt=0.25`. The results are marginally better without gil. before this PR  after this PR  Next,...

chunk size is 1024 in the previous examples

> I guess the flat 100 % in the beginning is the Satpy file handler creation we've already discussed in [pytroll/satpy#2186](https://github.com/pytroll/satpy/issues/2186) ? yes, that looks like it. Far to long...