DanRyanIrish

Hi @yashrsharma44. I'm not sure @Cadair needs to give feedback at this time. Let's push ahead with this plan!

The ```data```, ```uncertainty```, and ```mask``` can be primary/image HDUs. @Cadair suggested the ```extra_coords``` could be combined into a single Table HDU.

@hayesla and @grahamkerr request this functionality in #409. Their conversation has been redirected here. Although note that much of the above discussion was had in the context of ndcube 1.x....

That's a very good point. So in APE-14 land there is no generalized way to save to FITS? Given the scope of ndcube, that makes me unsure about a save...

What about HDF5? If gWCS can be serialized to that and we provide a reader as well, then maybe we don't need to provide a FITS writer.

Hi @dstansby. Thanks for raising this. I certainly see your point and appreciate this is a bit of a pain for EIS data. However, I have a philosophical concern for...

Ah I see. That is a different problem. Off the top of my head I don't know how to implement that but it must be possible...

@rosteen, would like to elaborate a little on this request?

Not sure which issue we're using for this discussion. But see my summary of our discussion at the STIX meeting and my current suggested approach [here](https://github.com/sunpy/sunxspex/issues/81#issuecomment-1806938942) on issue #81

Regarding point 2, I think the bigger thing is to decide on the API. Refactors for greater speed and/or Pythonic-ness can be done at any stage with little penalty so...