This is the mail archive of the cygwin mailing list for the Cygwin project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: mintty window gets tilde characters "pressed" in it

Am 14.12.2016 um 18:48 schrieb David Karr:
On Mon, Dec 12, 2016 at 10:18 AM, David Karr <> wrote:
Win7.  I believe my Cygwin version info is "2.5.2(0.297/5/3) 2016-06-23 14:29".

I've been annoyed for a while with a symptom in my mintty windows.  I
will be typing along, or not, and suddenly a "~" character gets
injected into the stream. It often is the only character at the
prompt, so I can erase it and type my command, but it's really
annoying when it appears while I'm typing. I was just reminded of it
now when it injected the "~" right before I pressed enter on a
"mkdir", so I had to then remove the dir I just created and recreate

Is this something that Cygwin can control?
I've determined what "instigates" this, but I find that it only mainly
affects mintty and somewhat affects emacs. I've never seen a symptom
in any other application.

I use a tool called "Windows Caffeine" that is used to keep the
display awake.  It sends a "f15 key up" event every 59 seconds.  That
is what is producing the "~" character in the mintty window.  I've
also seen a symptom from this in Emacs, but it's harder to produce.
It's not as simple as just having a text buffer in focus. I haven't
been able to reliably produce it in Emacs, but the mintty symptom is
very reproducible.
I've checked with Caffeine and traced mintty. It receives F15 key down, the key up. Caffeine claims to send key up only but that's not true. Please report this as a bug to Caffeine. X11 does not seem to handle F13...F24 at all because they don't appear on a typical PC keyboard (that's my assumption), so it doesn't occur in xterm. It generates a ~ only in terminal applications because a number of key escape sequences end with ~.

The key event from Caffeine can be distinguished from a real key event, because it does not include keyboard scancode information. So a workaround is possible in theory. I'm not sure though whether that's a good idea as there may be other applications that inject keys the same way, e.g. drivers of other input devices, like Braille.
Opinions and information are appreciated.


Problem reports:
Unsubscribe info:

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]