On 08/01/2017 01:54 PM, Achim Gratz wrote:
> Steven Penny writes:
>> Could you link to Davids post? I do not see it for this month or July, and I
>> dont seem to have trouble with numlock on or off, even with July 31 version.
> 
> She can't, this was a discussion on IRC.

The gist of the conversation and today's patch: when numlock is on,
'alt-5' behaves as expected; but when numlock is off, 'alt-5' was
generating two keypresses at once (the escape sequence for ALT+keypad5
as on Linux when you release keypad5, AND the windows behavior of
injecting a codepoint after releasing alt).  Corinna's patch means you
can no longer bind alt-keypad5 to do something magic (this is because
the cmd behavior is that alt-keypad sequences work regardless of numlock
state); but doesn't stop you from binding ctrl-keypad5 or shift-keypad5.
 Yes, it means one place where cygwin can't completely emulate Linux;
but the thought on IRC was that more users are probably familiar with
windows alt-sequence codepoint generation than they are with Linux
alt-keypress bindings, and that less than 1% of the population would
even notice the difference.

-- 
Eric Blake, Principal Software Engineer
Red Hat, Inc.           +1-919-301-3266
Virtualization:  qemu.org | libvirt.org