Put the AquaTk code added to pyshell.main in bpo-24801 where it should have gone originally. (My fault, ultimately.) See aqua_context.diff. This is a step in factoring main (currently about 170 lines) into a manageable number of function calls.

I will try to add a test that 1. creates an editor window with a 'fixed' root and mocked .right_click_event method, generates a right click, and checks the mock event handler; 2. creates a context menu, etc.

Serhiy, the docstring for .class_bind says "Bind to widgets with bindtag CLASSNAME". I have not and do not see 'bindtag' defined. Am I correct that it must refer to a tk widget class, rather than a python tkinter class, or subclass thereof?

"bindtag" is an arbitrary string. Usually this is a name of Tk widget class of or "all". bindtags() allows to retrieve or set a list of bindtags associated with a widget. By default they are: full path of a widget, name of Tk class of widgets, full path of a toplevel window, "all".

I am asking because I noticed that every editor windows gets the same set of bindings set and destroyed, (as in the other part of the bpo-24801 patch). Since events have a widget attribute, using bound methods with a self parameter is not necessary, so all the duplication seems redundant. But class_binding to Text is too broad. It there any way to register 'subclasses', similar to the way one can register style options as applying to a subset of, for instance, all Buttons?

I think you can just use bindtags().

button = Button(...)
bindtags = button.bindtags()
button.bindtags(bindtags[:1] + ['MyButton'] + bindtags[1:])
root.bind_class('MyButton', ...)

[The light dawns, as the pieces connect together] The pseudoclass 'MyButton' would then consist of all widgets with 'MyButton' inserted into its bindtags list. Similarly for 'Editor'. I presume that a) tk has the equivalent of a master bind dict mapping bindtags to dicts mapping event sequences strings to event handlers, and b) .bind_class(cname, evseq, func) does the equivalent of "tagdict.get(cname, {})[evseq] = func".

I also notice that the introspective versions of .bind... call will let us discover the pre-defined bindings on difference systems and check our alterations. Anyway, thanks. I will experiment when ready to refactor bindings. In the meanwhile, I will apply the patch and close this.

New changeset 09ec7f7322b5 by Terry Jan Reedy in branch 'default':
Issue bpo-27262: move Aqua unbinding code, which enable context menus, to maxosx.
https://hg.python.org/cpython/rev/09ec7f7322b5

I am leaving this open to look into later adding an automated test for this to the new test_editmenu file that will be added for bpo-5124.

New changeset 374dd14cf0e5 by Ned Deily in branch 'default':
Issue bpo-27262: fix missing parameter typo
https://hg.python.org/cpython/rev/374dd14cf0e5

I will add a test that calls setupApp, which would have failed with an error. I will also remove the call to _init_tk_type in setupApp, and let it be called when needed by the first isMacTk function called. Test code can partly simulate being on a particular type of Mac by setting _tk_type.

Changeset d8a2b6efdd4a, bpo-27239, adds a test of setupApp in test_macosx that fails on any system, in particular mine, when Ned's fix is undone.