On Thu, Sep 22, 2016 at 11:57:13AM -0300, Mauro Carvalho Chehab wrote:

Em Wed, 21 Sep 2016 10:54:21 +0100
Sean Young [off-list ref] escreveu:

quoted

We might want to know what protocol a remote uses when we do not know. With
this patch and another patch for v4l-utils (follows), you can do that with:

./ir-keytable  -p rc-5,nec,rc-6,jvc,sony,sanyo,sharp,xmp -t
Testing events. Please, press CTRL-C to abort.
1474415431.689685: event type EV_MSC(0x04): protocol = RC_TYPE_RC6_MCE
1474415431.689685: event type EV_MSC(0x04): scancode = 0x800f040e
1474415431.689685: event type EV_SYN(0x00).

This makes RC_TYPE_* part of the ABI. We also remove the enum rc_type,
since in input-event-codes.h we cannot not use enums.

In addition, now that the input layer knows the rc protocol and scancode,
at a later point we could add a feature where keymaps could be created
based on both protocol and scancode, not just scancode.

We need Dmitry's ack in order to apply this one.

I'd rather not: I am trying to keep input API hardware-independent and
the kind of device emitting keycodes (a remote control in the sense of
drivers/media/rc or USB device or BT device) should not really matter to
consumers. Similarly how we do not export whether device is USB1.1 or
USB2 or USB3 (although we do have input->id.bustype, but it is more for
identification purposes rather than for adjusting properties).

For configuration (like loading keymaps) we can examine
parent hardware device and decide.

Thanks.

-- 
Dmitry