nocturnalastro
commented
4 years ago Thanks from a quick look I don't think we're effected to much, although I have next week off to do astro stuff, I'll sit down dig in.
When looking as some local failures, I noticed that the evaluate function for CompoundModels was just a pass as this the function at the core of saba this means. We can not work with CompoundModels. I opened a PR. The most of the code in the docs will fail as I often used two Gaussians for the model in the docs
In v4.0 astropy has done some major API changes in modelling.
http://docs.astropy.org/en/stable/whatsnew/4.0.html#whatsnew-4-0-modeling-api
http://docs.astropy.org/en/stable/modeling/changes_for_4.html#modeling-major-changes-for-4-0
We need to review how this impacts saba. Of course, we should always aim for backwards compatibility is possible, but if it turns out that it is difficult to support both astropy < 4.0 and >- 4.0, but I argue that saba could make the leap and require astropy >= 4.0. For now on.
@nocturnalastro @taldcroft