yjftsjthsd-g / keynav Goto Github PK

Sometimes I thought that keynav works wrong and scrolling, clicking nums 1-3 don't work. But accidentally it start work. Long time I cant understand when it goes right and whet it's not. But I found the answer (may be not full):

I have two displays in my setup. When I start keynav grid on main display "click" action works well in right position of cursor. But if I started keynav on second display, it doesn't work. If I move grid with shift to the first screen and after try use "click" it will work correctly.

Hope it helps.

The current manpage contains the following:

SEE ALSO
       Related: xdotool(1)

       Project site: <http://www.semicomplete.com/projects/keynav>

       Google Code: <http://semicomplete.googlecode.com/>

CONTACT
       Please send questions and comments to [email protected]. File bugs and feature requests at the following URL:

       <http://code.google.com/p/semicomplete/issues/list>

Minimally, Google Code is dead so that should be pointed to this repo. Need to check whether that mailing list still exists; probably best to suggest opening an issue on this repo.

Currently, compiling on GNU/Linux systems throws, "keynav.c:376:9: warning: implicit declaration of function 'asprintf'". This does not break the build and does not appear on FreeBSD. It appears that asprintf is a GNU extension also implemented by some or all BSDs. There are two ways to resolve this: We can either 1. refactor the code to not use asprintf (it appears that its functionality can be replaced with calloc+sprintf), or 2. just say that we're using GNU extensions (this is done by one other keynav fork; see dradetsky@247d1a6)

Currently, the Makefile generates a binary that is not stripped and apparently has some level of debugging info present, but isn't built with -g or anything. There are commented-out flags for both adding more debugging info to the binary and for building with -O2, but none of them are active.
This issue is to add release and debug targets that will respectively build a stripped and optimized binary, or one with debugging explicitly supported. One possible method is discussed on this Stack Overflow post but it is unclear that this is portable or specific to GNU Make. Bonus points for being able to build a static release binary.

Building with CFLAGS+=-Wall produces a full screen of warnings, mostly apparently unused variables. Each warning should be evaluated to determine (if possible) why the warning is happening, what the code in question is intended to do, and to fix the warning.

It would be quite nice if we could automatically test keynav, mostly so we can make changes and make sure something didn't break. This is a little annoying because it's an X11 program that has nearly no GUI:)

Rough plan:

• run headless X server (xvfb?)
• provide simulated keypresses
• check the cursor position
• use xev to listen for events like mouse clicks

Of course, once that's working someone will have to actually figure out what to test:)