This is not difficult to do and quite a lot of people will ask for it, so let's aim for this to be done for the Beta release or soon afterwards.

Let's define how this should work. I can see it working in two different ways:

• You have a Marvin Cube cube and you do cube.get_all_maps() and get a list of Marvin Maps objects already instantiated. This is convenient but over the API it may be a bit slow.
• You do cube.list_available_maps() and get a list of string or ad-hoc object that basically tell you the bintype and stellar library combinations available. You could yo something like

>>> av_maps = cube.list_available_maps()
>>> av_maps
['ALL-GAU-MILESHC', 'SPX-GAU-MILESHC']
>>> type(av_maps[0])
MarvinClassNameTBD
>>> maps = av_maps[0].load()

Comments?

I like the 2nd option that provides a simple string list of what is available. Then the user can decide what they want to instantiate. Option 1 might be too slow and demanding on the server.

I was just thinking of a string list, and we may not need anything fancier. They could do

>>> av_maps = cube.list_available_maps()
>>> av_maps
['ALL-GAU-MILESHC', 'SPX-GAU-MILESHC']
>>> cube.getMaps(avmaps[0])

but your way could work too.

I like that. It would require reworking the cube.getMaps a bit but it should not be too hard, and it's useful for getMaps itself. I'll implement that.

Another subtle issue. Should we provide a list of all possible maps (that would always be the same list of 4 maps for MPL-5 and 9 for MPL-4) and use the normal loading logic in Maps (i.e., first will try to use the same mode as the cube, but if it fails it will revert to remote mode); or should we only show the list of maps available for that mode. The latter would mean that if you open a cube from a file it would only show the maps for which you have the file, and the same with the DB.

I like the idea of only showing the list of maps that are actually available. In principle, it should be all of them anyways, at least remotely, but doesn't have to be in local mode. But then that does require you to always search the filesystem for those maps. I guess it's not so bad if we restrict it to the maps location in the SAS directory.