Panu Lahtinen

> * Other solutions? I think the easiest route, at least for my own worklfow, is to pre-resample the background images (black marble and land-sea mask) to my target projection...

@simonrp84 so `scn[chan].attrs['area'] == scn2['image'].attrs['area']` is `True`?

> I don't have a problem waiting a little longer, but don't all the readers use hdf5plugin right now? We don't import `hdf5plugins` in Satpy. It was decided that we...

Geotiff can hold "many" bands. In a project I've created 7-band files using `gdal_translate`. Unfortunately I don't have the code at hand, but IIRC it was pretty simple. Don't know...

I've been trying to speed-up remote FCI reading by doing some caching in `h5netcdf` (https://github.com/h5netcdf/h5netcdf/pull/221), and now the biggest bottleneck seems to be that there are gazillion requests made to...

Forgot to link my more recent test with parallel filehandler for FCI, so for future reference: https://pytroll.slack.com/archives/C011NR3LE20/p1699870659317759 There's also link to the branch I used. So the parallelization works but...

One workaround in tests seems to be to do an `arr.compute()` and use `max_computes=1`. In #2623, before the fix, the scheduler reported four computations when triggered. So cause one intended...

Because I haven't seen this with FCI, using `scn.load(..., generate=False)` should be a valid workaround (it is generally also much faster if the channels are used in multiple composites). I...

@djhoese my recollection from the previous inspection was that the `match_data_arrays()` didn't `raise`, but it seems that part is further down in `check_geolocation()`. And the ValueError would've been the one...

This happens because gradient search defaults to bilinear interpolation and thus changes the original values used to create the mask to floating point. Switching to nearest should resolve this: `scn.resample(...,...