David Kastrup <dak@gnu.org> writes:

> Richard Stallman <rms@gnu.org> writes:
>
>>     > If you thing it reasonable that key-binding should also react to
>>     > bona-fide events _not_ wrapped in an array, I can do this easily
>>     > enough in the change I am working on.
>>
>> Please don't make that change.  key-binding is supposed to look up a
>> sequence of events; it expects that sequence to be in a vector or
>> string.  There is nothing wrong with that, and no need to change it.
>>
>> I think we simply had a misunderstanding about what it means to say
>> that key-binding can "look up events".  In my way of looking at
>> things, you give key-binding a vector of events, and what it does is
>> "look up events."
>
> Ok, I'll rework my patch.

Ok, here is the complete patch.  I also added an optional argument to
`command-remapping' (this was required to let `key-binding' do the
right thing in the case of mouse events).

I removed the `kill-emacs' call if the current frame not live: I can't
think of it making much sense in `key-binding'.

I also added a changelog entry and tested this with a number of
events.  Looks clean to me.

I would like to check this in soon so that one can make the mouse
remapping stuff use it.  That should make the `follow-link' property
work the same as any remapping.