Re: [PATCH] Override Windows default Win-* key combinations when using Emacs

[Jussi Lahdenniemi]

> > The patch is attached. I decided to place the new functionality in a new
> > file (w32hook.c) - it could have been added to some existing file as well.
>
> I'd prefer not to create a new file, but to have this in w32fns.c.  It
> might even allow you to make some functions static.

OK.

> > address@hidden w32-register-hot-key
> > address@hidden w32-unregister-hot-key
> > -  MS-Windows reserves certain key combinations, such as
>
> I'm not sure I understand why you moved this stuff down in the file.
> Was something wrong with its original place?  You could make the
> changes without moving the stuff, no?

In the documentation for w32-register-hot-key I am now referring to the 
w32-[lr]window-modifier variables; there is a strongish dependency 
between these now. For this reason, I preferred documenting the 
w32-*-hot-key functions after the w32-*-modifier part. But of course, 
this is not a must.

> > +   The hotkey registrations always also include all the shift and
> > +control modifier combinations for the given hotkey; that is,
> > +registering @address@hidden as a hotkey gives you @address@hidden,
> > address@hidden@key{a}} and @address@hidden as well.
>
> I believe you meant to use "C-S-" etc. here, for the Shift key (not
> "super").

No, what I mean is that registering s-a also registers S-s-a, C-s-a, and 
C-S-s-a. I wrote S-*-a as *-A in that documentation, base on the 
observation that for example C-h k M-S-a is shown as M-A by Emacs. But 
maybe it would be better to write a lower-case a and explicit S modifier 
in all the cases?

> > -                 if (NUMBERP (Vw32_phantom_key_code))
> > -                   key = XUINT (Vw32_phantom_key_code) & 255;
> > -                 else
> > -                   key = VK_SPACE;
>
> It looks like you are removing the support for this feature.  Is it
> really not needed anymore?  If so, we should mention this change in
> NEWS, and tell there how to achieve the same effect without this
> variable.

True. I have never used the w32-phantom-key-code myself, so I did not 
notice that (and, actually, the variable is still checked - 
unnecessarily - in w32inevt.c). But, setting 
w32-pass-[lr]window-to-system to NIL has now the desired effect even 
without the use of a phantom key code; the keypress is routed directly 
from the hook function to the Emacs input handler, without passing 
through the standard system keyboard functionality inbetween, which 
eliminates the need to lie to the system about the key that has been 
pressed.

> I've briefly reviewed the rest of the patch, and it looks OK to me.
> Did you make sure the new code doesn't call any API that might be
> unavailable on older systems?  We still support Windows 98 and later.

Unfortunately, Windows 98 does not include the low-level keyboard hook 
functionality that is essential for my code to work. It is supported in 
Windows 2000 (and even NT4, apparently), but not on the 95/98/ME branch.

So, probably I would need to change the code so that the old 
functionality is preserved and used if the new API are not available 
(pre-Win2000), in the same style that w32uniscribe works? It should 
still even be possible to write a single configuration file that works 
on both environments.

> Also, I understand that this code will work on versions of Windows
> before v8, is that right?

I have tested the code on Windows 10 (my physical PC) and Windows 7 (in 
a VM) both as 32-bit and 64-bit versions. I'll try to get Win2000 
installed on another VM as well and test with it, just to be sure.

--
Jussi Lahdenniemi