Auto key repeat in wayland.

[Scott Moreau]

On Sun, Mar 17, 2013 at 7:47 PM, Yichao Yu <yyc1992 at gmail.com> wrote:
> Hi Scott,
>
> On Sun, Mar 17, 2013 at 8:51 PM, Scott Moreau <oreaus at gmail.com> wrote:
>> On Sun, Mar 17, 2013 at 5:36 PM, Yichao Yu <yyc1992 at gmail.com> wrote:
>>> Hi,
>>>
>>> Seems that weston uses a client side keyboard auto-repeat that is
>>> HARD-CODED in `window.c` and I haven't seen anywhere in wayland's
>>> mentioning how this should be done in wayland. There doesn't seem to
>>> be a place where the client can get a recommended repeat delay and
>>> frequency from the server either. Does it mean each program/toolkit
>>> will then have it's own key repeat setting or even hard coded in the
>>> code that cannot be change without a recompiling (and in either case
>>> it is not possible to set it globally anymore)???
>>>
>>> Why can't this be done in the server or shouldn't a least a
>>> recommended value that can also be set from client be added?
>>>
>>> Yichao Yu
>>
>> Hi Yichao,
>>
>> I have asked about this several months ago and the decision was made
>
> This one[1]?
>
>> by wayland developers that each toolkit should implement this feature
>> if they would like to have it. My position is that the key repeat
>
> And the decision was all toolkits/programs written without using a
> toolkit ALL need to "agree on where to retrieve that (key repeat
> configuration) from"??? Shouldn't that "where" be the compositor??
>
>> events should originate from compositor side. Not only would this give
>> this expected behavior of a consistent key repeat timing and always be
>> available by default, but it would let us have things such as key
>> repeat for key bindings in the compositor. Martin also expressed that
>> even the Linux kernel has key repeat functionality. Here is a quote
>> from Martin on IRC:
>>
>> <min2_> there is a key repeat functionality in kernel btw, no need to
>> recreate in toytoolkit
>>
>> There's no reason this should be a client-side implementation. It is
>> hard to get wrong in a single implementation but having a desktop with
>
> At least clutter and weston both have a hard coded key-repeat setting
> (and seems to be different....), which IMHO is "get wrong"......
>
>> each client supporting it differently or not at all is just
>> ridiculous. The sheer code and effort duplication that would be
>> required is entirely unacceptable. I don't think this is an option, I
>> think this is a requirement of any sane display server for a desktop
>> platform.
>
> So is the patch just disappear in nowhere? IMHO, it will also be very
> useful to make it possible for a client to change that setting
> (instead of a hard coded value in the compositor or a value in a
> configure file that require a restart to take effect).
>
> OT: That patch seems to have a set of hard coded non-repeat keys.
> Isn't this setting (which key should be repeat) part of xkb? (Anyway,
> since the X server can figure that out from keyboard layout, whatever
> backend wayland use shouldn't have any problem on that.....)
>
> Yichao Yu.
>
>>
>> - Scott
>
> [1] http://lists.freedesktop.org/archives/wayland-devel/2012-March/002511.html

Hi Yichao,

Thanks so much for digging up this patch, I had forgotten all about
it. I've started work in the next branches on github so I'll apply it
there and get rid of toytoolkit repeat.

- Scott