Re: [PATCH v3 05/18] oob: Allow to register pairing complete callback

On Monday 24 of September 2012 12:06:26 Johan Hedberg wrote:

Hi Szymon,

Hi Johan,

On Fri, Sep 21, 2012, Szymon Janc wrote:

quoted

--- a/src/mgmt.c
+++ b/src/mgmt.c

@@ -1193,6 +1193,7 @@ static void pair_device_complete(int sk, uint16_t index, uint8_t status,
 	info = &controllers[index];
 
 	bonding_complete(info, &rp->addr.bdaddr, status);
+	oob_pairing_complete(&info->bdaddr, &rp->addr.bdaddr, status);
 }

Can't we be a bit smarter than always notifying the oob code. Don't we
e.g. have the possibility to look at the IO capabilities used in the
pairing process or some other means from which we can conclude that this
was an OOB-based pairing that completed?

Hash and randomizer are optional tags (only address is mandatory) so there
might be only OOB discovery and in-band pairing (OOB data present flag not
set in iocapa).

I could try to make OOB callbacks registering mechanism a bit more runtime
aware (register/unregister on-demand basis) or bit-more compile time aware
(hide after a macro and enable only if one of oob plugin is build).

Not sure if this would not over-complicate that code to much... opinions?

Johan

-- 
BR
Szymon Janc