Device Properties Protocol Spec - Draft 1

[Peter Hutterer]

On Mon, Sep 08, 2008 at 11:19:21AM +0200, Simon Thum wrote:
> Peter Hutterer wrote:
>> The difference to the current implementation is that there is no
>> meta-information about the property. This information included immutable,
>> client-created, range and pending. I deem all four as too much information.
>>
>> If such information is required in the future, such requests can be added
>> easily in future versions of XInput.
> I guess you are just not exposing those?
> My concern is that a client could remove properties installed by the  
> server or a driver.
>
> In the readonly case, I think it should be possible to tell a property  
> which refuses to be set due to the value apart from a property which  
> generally is 'read-only', i.e. has a always-fail setter or some  
> unexposed flag. Could be BadValue vs. BadMatch, but I'm missing it in  
> the ChangeDeviceProperty spec (readonly is arguably a type mismatch).

we could do that as BadAccess, though I'm not sure how that'd mingle with the
security stuff.

Cheers,
  Peter