allow multiple instances of a Magic

[minrk]

as reported on-list, Magics were using a class magics mapping, bound to the first instance, rather than creating a mapping for each instance.

This may have been by design, but it offers little-to-no benefit, and surprising unchangeable behavior in (admittedly unlikely) cases.

[fperez]

More than by design, it was by expedience in the rush of getting everything done. Definitely this needs a cleanup, I've already thought of other cases where the current implementation will give us headaches.

[minrk]

Well, given all the cell magics people have been contributed, it's clearly highly valuable and great work. But it is pretty complex, so we are going to find little weird bugs like this while we are putting it through the ringer.

[ivanov]

mark this with @dec.skip_known_failure to just merge and move on, or is the intent for this to be a work-in-progress?

[minrk]

oops, no - there should be two commits, test and fix. I just forgot to push the second :)

[minrk]

Test results for commit 139d6c0 (can't merge cleanly)
Platform: darwin

• python2.6: OK (libraries not available: cython matplotlib oct2py pygments pymongo qt rpy2 tornado wx wx.aui)
• python2.7: OK (libraries not available: oct2py wx wx.aui)
• python3.2: OK (libraries not available: cython matplotlib oct2py pymongo qt rpy2 wx wx.aui)

Not available for testing: python3.1

[minrk]

Any reason this shouldn't be merged? It's relatively minor, so I would be okay letting it slip to after 0.13, but the side effects of the bug are quite surprising when they come up (as seen on-list).

[fperez]

No, no reason, just too busy. I'd had a first look before and just didn't finish, sorry.

This is a good fix for a wart from the cell magics work, so it should go in. Thanks for the good work!

Merging now.