Re: [PATCH 2/2] git-gui: add hotkey to toggle "Amend Last Commit" check button/menu

From: Pratyush Yadav <me@yadavpratyush.com>
Date: Sep 13, 2019, 17:47 UTC
Message-ID: <20190913174743.x2kdfotq2fnuift7@yadavpratyush.com>
In-Reply-To: <CAGr--=JDKrYT=Z8iwPVRy58OB9teH7bHXrYSZHqYHhJ=5wrJzQ@mail.gmail.com>

On 13/09/19 09:50AM, Birger Skogeng Pedersen wrote:

Show 22 quoted lines

> Hi Marc and Philip,
> 
> 
> On 12/09/2019 22:34, Marc Branchaud wrote:
> > I disagree!  Who expects anything to work properly when capslock is on?
> 
> Me :-)
> 
> 
> On Fri, Sep 13, 2019 at 12:23 AM Philip Oakley <philipoakley@iee.email> wrote:
> > I'd tend to agree. In other areas the use of shift is often used as the
> > complement of the unshifted action, so it does feel 'odd'. Thus it could
> > be used directly as the bool for amend or direct commit.
> >
> > This all assumes that Caps Lock is equivalent to having the shift on,
> > rather than being a special extra key.
> 
> It seems all the Ctrl+(lowercase character) hotkeys in git-gui have an
> equivalent Ctrl+(uppercase character).
> So for this feature, we should keep the Ctrl+E bind aswell as the
> Ctrl+e bind. If nothing else, to keep it consistent with the rest of
> the hotkey bindings.

I agree with this that we should keep it consistent with the rest of the bindings for now...

> But honestly, (as Marc pointed out) it is a quite weird that
> Ctrl+Shift+(character) has the excact same function as
> Ctrl+(character). Perhaps we should find another way to bind the
> hotkeys, where the state of Caps Lock doesn't matter? If possible.

...but I'd love to see this happen. To me shift is a modifier. No matter whether Caps Lock is pressed or not, it should not do the shift-modified behavior (that's just me, maybe other people think differently).

AFAIK, Tk does not provide any direct way to find out whether shift is pressed (correct me if I'm wrong). What you instead have to do is some bit arithmetic on the number passed to the "Key" event via the "%s" substitution. Source: [0]. We can probably have a bind_alpha procedure that takes two arguments: what to run when shift is pressed and what to run when it isn't.

This, of course, would be incompatible with the current behavior, but do people even keep the Caps Lock on? I personally use it so rarely I have my Caps Lock bound to Escape because I might as well use that key for something I use more often.

[0] https://blog.tcl.tk/4238

-- 
Regards,
Pratyush Yadav