Atreyee Sinha

Handling such models in `gammapy.estimators` and `FluxPointsDataset` seems to be buggy. We only use `Model.spectral_model.evaluate` and miss the units (eg: https://github.com/gammapy/gammapy/blob/bf20ed4c6549924b3f79ac8f978bdebeae1e9d65/gammapy/datasets/flux_points.py#L269) ### To reproduce ``` from astropy import units as...

I don't have a better solution for the moment ... but must admit it is annoying for a LTS version, because this can be useful for studies of the the...

@adonath Maybe re-label this to `1.1` or later and keep open as a reminder? This is something we should support in future

@maxnoe > Why is it not possible? This is not possible because of the lazy loading. If you explicitly try to set `obs.events = events_new` it raises `AttributeError: can't set...

In general, I think the behaviour of `Observation.create` is not proper as it does not use the `rad_max` irf https://github.com/gammapy/gammapy/blob/89f67e3014c3673704ea6eeb8ad3ad0b844a3426/gammapy/data/observations.py#L223 I am not putting a milestone now, but I think...

Okay this is a bit strange It seems the behaviour is different between `Observation.read()` and `Datastore.obs()`

I think it would be better to edit the docs section once we have a final decision. For now, I just summarise the present status here. 1.For 1D spectral analysis...

Thinking about this, I tend to agree with @registerrier I started with (1) to be analogous with `SpatialModels` - you **always** need to multiply by a `SpectralModel` for the normalisation...

Closing this PR and opening a new one

Looking at the document, I see that the suggestion is to use `tableau- coloblind10` for colorcycles, and `viridis` (or `cividis`) for colormaps. Does this have to be done for each...