On Sat, Apr 22, 2017 at 10:16:13AM -0700, will sanfilippo wrote: > Again, sorry if I missed something here or it has been discussed. This would > only work with a device that would accept this and not all do but there is > the connection parameters request procedure that a slave can use to change > the supervision timeout being used. The master can also use this procedure or > it can use the connection update procedure. This way you do not have to > change the connection interval. Not sure what the accessory design guidelines > say about doing this. As Chris mentions, mynewt does not do this currently > but it could when doing an image upgrade.
If either controller does not support the link layer Connection Parameters Request procedure, the slave host will send a L2CAP Connect Parameter Update Request instead. The L2CAP procedure is a slightly weakened version of the link layer procedure. So, some form of connection update procedure will be supported in all connections. I was thinking the slave would be helpless to update the connection unless both devices support the link layer procedure, and almost expressed the same warning yesterday, but then I remembered the L2CAP procedure. So I guess I am injecting my forgetfulness into your response! Sorry If I am misinterpreting your words. > > The min_ce_len and max_ce_len, btw, are not really used by the Mynewt > Controller. Not sure what other controllers do with this information but it > is meant to control how long the connection event lasts. As Chris mentions, > it is not exchanged between peers and is just a “request” from the host to > the local controller which the controller does not have to honor. Not really > useful in this context anyway. Thanks for clarifying that. Chris
