[PATCH BlueZ 0/2] Change ConnectDevice to optionally connect services

[PATCH BlueZ 0/2] Change ConnectDevice to optionally connect services

[Karabin, Josh]

This patch set makes it possible to use the experimental Adapter1
ConnectDevice DBUS API to connect to a device without also
connecting each of the devices service. This enables out-of-band
discovery of an address with the conditional behavior of calling
ConnectProfile on a discovered device.

I chose to change the default behavior of the ConnectDevice API.
It will only connect all services if you set a new ConnectAllDevices
boolean value in the dictionary argument.

I feel that that makes a better default, and with this still being
an experimental API, I'll ask to see how others feel.

This would require changes to users of the original commit:

commit c3f7b6733c5f ("adapter: Add ConnectDevice method")

I would be fine with changing the API in some other way that makes this
use case possible - this is just my preference.

You may have seen questions about this a little bit ago - this is the
follow-up:

https://marc.info/?l=3Dlinux-bluetooth&m=3D153317507925739&w=3D2
https://marc.info/?l=3Dlinux-bluetooth&m=3D153315011619515&w=3D2

George Josh Karabin (2):
  Change the experimental Adapter1 ConnectDevice api to default to not
    connecting all profiles, using an optional parameter to restore the
    current behavior.
  Update the Adapter1 API to match the change to ConnectDevice.

 doc/adapter-api.txt | 15 ++++++++++-----
 src/adapter.c       | 18 +++++++++++++++---
 2 files changed, 25 insertions(+), 8 deletions(-)

--=20
2.17.1

Re: [PATCH BlueZ 0/2] Change ConnectDevice to optionally connect services

[Luiz Augusto von Dentz]

Hi Josh,

On Thu, Aug 16, 2018 at 11:26 PM, Karabin, Josh
<George.Karabin@honeywell.com> wrote:
> This patch set makes it possible to use the experimental Adapter1
> ConnectDevice DBUS API to connect to a device without also
> connecting each of the devices service. This enables out-of-band
> discovery of an address with the conditional behavior of calling
> ConnectProfile on a discovered device.
>
> I chose to change the default behavior of the ConnectDevice API.
> It will only connect all services if you set a new ConnectAllDevices
> boolean value in the dictionary argument.
>
> I feel that that makes a better default, and with this still being
> an experimental API, I'll ask to see how others feel.

I wonder how this is suppose to work since what you are suggesting is
to not auto-connect matching profiles/services, but that would leave
the ACL link with no channels in use (at least on BR/EDR) so the link
would disconnect anyway.


> This would require changes to users of the original commit:
>
> commit c3f7b6733c5f ("adapter: Add ConnectDevice method")
>
> I would be fine with changing the API in some other way that makes this
> use case possible - this is just my preference.
>
> You may have seen questions about this a little bit ago - this is the
> follow-up:
>
> https://marc.info/?l=linux-bluetooth&m=153317507925739&w=2
> https://marc.info/?l=linux-bluetooth&m=153315011619515&w=2
>
> George Josh Karabin (2):
>   Change the experimental Adapter1 ConnectDevice api to default to not
>     connecting all profiles, using an optional parameter to restore the
>     current behavior.
>   Update the Adapter1 API to match the change to ConnectDevice.
>
>  doc/adapter-api.txt | 15 ++++++++++-----
>  src/adapter.c       | 18 +++++++++++++++---
>  2 files changed, 25 insertions(+), 8 deletions(-)
>
> --
> 2.17.1
>



-- 
Luiz Augusto von Dentz

RE: [External] Re: [PATCH BlueZ 0/2] Change ConnectDevice to optionally connect services

[Karabin, Josh]
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=

Re: [External] Re: [PATCH BlueZ 0/2] Change ConnectDevice to optionally connect services

[Luiz Augusto von Dentz]

Hi Josh,

On Fri, Aug 17, 2018 at 4:35 PM, Karabin, Josh
<George.Karabin@honeywell.com> wrote:
>> -----Original Message-----
>> From: Luiz Augusto von Dentz [mailto:luiz.dentz@gmail.com]
>> Sent: Friday, August 17, 2018 2:15 AM
>
>> I wonder how this is suppose to work since what you are suggesting is
>> to not auto-connect matching profiles/services, but that would leave
>> the ACL link with no channels in use (at least on BR/EDR) so the link
>> would disconnect anyway.
>
> In my use case, the client application will wait for the appearance of
> a device object with a matching address to appear on the Device1
> interface. If that device contains a profile that that the application
> supports, it will call ConnectProfile on the device object. In general,
> the link won't time out before the application has a chance to
> respond, but it would be fine if it did - the ConnectProfile call would
> end up re-establishing the link in that case.
>
> I could note that behavior in the adapter api document if you think
> It would help to clarify the intent.

Wouldn't it be better to have a list of profiles to connect then? That
would make a nice addition to the API since ConnectProfile only
connect one profile at time, that way we could have auto-connect be a
list and in case it is empty we connect all otherwise we just connect
the profiles in that list, how about that?

-- 
Luiz Augusto von Dentz

RE: [External] Re: [PATCH BlueZ 0/2] Change ConnectDevice to optionally connect services

[Karabin, Josh]
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