Re: [gtk-vnc-devel] PATCH: Fix scancode mapping for evdev

[Anthony Liguori <anthony codemonkey ws>]

Daniel P. Berrange wrote:
> The VNC extension for sending raw scancodes assumes that the scancodes 
> map to the keycodes with xfree86 layout. This doesn't hold true when
> you introduce evdev to the world. Easiest demonstration of the problem
> is to press the 'Up' key, and see 'PrintScreen' arrive in the guest.
>
> This patch adds a new mapping table with the evdev data. Although
> Ubuntu have a patch like this in their gtkvnc packages for months
>   

It hasn't been months, it's been weeks :-)

> (never been submitted back here), it copies from the GPL licensed
> VirtualBox - you can argue whether data tables are exempt from this
> or not, but I feel its better to not copy it in the first place
> since we can generate it easily enough. It also re-probes for whether
> its using evdev on every single key press event which is creating 
> far too much X traffic.
>   

FWIW, I wrote that patch and I haven't submitted it because of the 
website issues.  The table is from VirtualBox but I checked the table in 
the evdev keymap and all of the data (the keycodes and the comments) 
come from that file.  There are a couple of VirtualBox fixes.  Since it 
is trivial to write a program to generate the exact same table from the 
evdev source file, I don't believe VirtualBox can claim to hold 
copyright on it.  Ignoring that, such a table certainly is certainly 
considered a fact and therefore not copyrightable.

The check_for_evdev() probe should definitely be cached.

I'm really indifferent to what approach we use.  They're equivalent as 
far as I can tell.  I just wanted to provide a little explanation behind 
the Ubuntu patch :-)

Regards,

Anthony Liguori