All of lore.kernel.org
 help / color / mirror / Atom feed
From: Jakub Pawlowski <jpawlowski@google.com>
To: Marcel Holtmann <marcel@holtmann.org>
Cc: Luiz Augusto von Dentz <luiz.dentz@gmail.com>,
	"linux-bluetooth@vger.kernel.org"
	<linux-bluetooth@vger.kernel.org>,
	Arman Uguray <armansito@chromium.org>
Subject: Re: [PATCH BlueZ v5] doc/adapter-api.txt: StartFilteredDiscovery method.
Date: Tue, 24 Feb 2015 15:12:08 -0800	[thread overview]
Message-ID: <CAD0=r8yw=6bNJDK1YcK4=MpBpQ=mYKpaxKKJHSFeyajAuZ_Cjw@mail.gmail.com> (raw)
In-Reply-To: <D54A2499-C0C9-4EC6-8550-DBACD6210842@holtmann.org>

On Tue, Feb 24, 2015 at 2:50 PM, Marcel Holtmann <marcel@holtmann.org> wrot=
e:
> Hi Jakub,
>
>>>>> Since we can't any more track multiple different discoveries within t=
he
>>>>> same application (D-Bus connection) the StopDiscovery behavior is now
>>>>> quite broken. The simplest way around that would be to add a discover=
y
>>>>> instance return parameter to StartServiceDiscovery and to have a new
>>>>> StopServiceDiscovery D-Bus method that'd take this as an input
>>>>> parameter.
>>>>
>>>> Right now each application can call StartDiscovery only once, I want
>>>> each application to be able to call StartDiscovery OR
>>>> StartFilteredDiscovery only once. This way we are completly fine with
>>>> one StopDiscovery method. If some application needs to change filter,
>>>> it can always stop and restart it's scan.
>>>
>>> Actually I remembered wrong how the current code works. I thought it wa=
s
>>> allowing multiple StartDiscovery() per app, but there's indeed a
>>> hard-coded limit of just one. Keeping this policy around would allow
>>> StopDiscovery to be used for StartDiscovery and StartServiceDiscovery
>>> alike, and each app would then only be allowed to call one of those at =
a
>>> time.
>>>
>>
>> So I want to keep the limit of one session per app, but I think that
>> app should be able to call StartServiceDiscovery multiple times, in
>> order to update it's filter without calling StopDiscovery to first
>> stop it's discovery. That would simplify filter updating for apps that
>> would have such need, and make their logic simpler.
>
> we have never done such an API design before. I am a bit reluctant to all=
ow that now. The whole API design is falling into one simple approach that =
the name of method calls and properties is pretty self explanatory. That St=
artServiceDiscovery means also additionally UpdateServiceDiscovery is not s=
omething I would consider intuitive.
>
>> Example: App wants to update it's filter. If It can't call
>> StartServiceDiscovery more than once, here are steps:
>> 1. app call StopDiscovery.
>> 2. Discovering property changed to false.
>> 3. It calls StartFilteredDiscovery with updated filter.
>> 4. Discovering property changed to true
>>
>> If calling StartFilteredDiscovery more than once is possible:
>> 1. app call StartServiceDiscovery with updated filter, filter is
>> updated internally and "Discovering" property is not updated at all.
>>
>>
>> I think I argree with all other comments from other emails. I'll
>> update documentation accordingly.
>> I'll say in doc, that StartServiceDiscovery might be called multiple
>> times, and keep only one session. If anyone doesn't argree with that
>> I'll remove that, as that's only for convinience, and doesn't really
>> affect functionality.
>
> While I just looked at this for only a few minutes, the double execution =
of StartServiceDiscovery is something that does not fell right. Why not int=
roduce UpdateServiceDiscovery or ChangeServiceDiscovery.

Ok, I'll make it so you can call StartFilteredDiscovery only once. If
you want to change filter just stop and start the filtered scan again.

>
> Regards
>
> Marcel
>

      reply	other threads:[~2015-02-24 23:12 UTC|newest]

Thread overview: 15+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-02-10  0:34 [PATCH BlueZ v5] doc/adapter-api.txt: StartFilteredDiscovery method Jakub Pawlowski
2015-02-16  8:46 ` Luiz Augusto von Dentz
2015-02-20 17:09   ` Jakub Pawlowski
2015-02-23 13:09   ` Johan Hedberg
2015-02-23 16:41     ` Jakub Pawlowski
2015-02-23 18:43       ` Jakub Pawlowski
2015-02-23 21:43         ` Arman Uguray
2015-02-23 22:44           ` Jakub Pawlowski
2015-02-23 23:07             ` Arman Uguray
2015-02-24  0:01               ` Jakub Pawlowski
2015-02-24 10:51                 ` Johan Hedberg
2015-02-24 10:37       ` Johan Hedberg
2015-02-24 20:23         ` Jakub Pawlowski
2015-02-24 22:50           ` Marcel Holtmann
2015-02-24 23:12             ` Jakub Pawlowski [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to='CAD0=r8yw=6bNJDK1YcK4=MpBpQ=mYKpaxKKJHSFeyajAuZ_Cjw@mail.gmail.com' \
    --to=jpawlowski@google.com \
    --cc=armansito@chromium.org \
    --cc=linux-bluetooth@vger.kernel.org \
    --cc=luiz.dentz@gmail.com \
    --cc=marcel@holtmann.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.