Matt Craig

Just to be clear, do you mean that it currently returns an `HDUList` but should return an appropriate `HDU`, or that it returns an `HDU` and should return a list?

Could this be an alias to fitsheader (already included in astropy core)? Or maybe a table in the docs like: | To do this IRAF task... | ...do this from...

About 4.5 minutes, but it feels longer..

Sure; I can also provide the data set I was working with if you want. It is just a bunch of dark frames. It is conceivable the bottle neck is...

Very related, thanks for mentioning that, @SaOgaz. And, of course, thanks to @larrybradley for working on it!

I haven't had a time to do much profiling, but here is an example of a set of calls that takes several minutes to run (for the combine part). Adding...

I've a had chance to do a bit of profiling; the notebook I used to generate the profiles is [in this gist](https://gist.github.com/mwcraig/5fbbcfd70c95c03fa3ebdf5675a6d12b). The notebook combines 25 images, each roughly 2k...

@MSeifert04 -- do you have time to follow up on this or should I take a look later today?

@RLWagner -- sorry for the delay. I'm not able to reproduce the bug, but I'm not on Ubuntu (not that it should make any difference). Could you please send (or...

There is no objection in principle to supporting that; a little more detail and/or an example would be helpful. Also, you may want to take a quick look at the...