[SailfishDevel] BlueZ upstreaming?

[Bob Summerwill]

I brought up BlueZ4 "harvesting" for Sailfish in this week's Mer Meeting:


http://merproject.org/meetings/mer-meeting/2014/mer-meeting.2014-09-09-15.00.html

Calling Hannu Mallat!   I am told that you are "the guy" :-)

I am forwarding this earlier e-mail thread to the Sailfish development
list, for archiving of the conversation, and so that Bluetooth developers
on Mer/Sailfish/Jolla can have this additional context.   Thanks!


Cheers,
Bob

On Fri, Aug 29, 2014 at 4:22 AM, Bob Summerwill <bob at summerwill.net> wrote:

> Thanks for the great summary, Javier.
>
> I just realized that I never pulled in the authors who Mikko highlighted
> earlier in the chain, so I will do so now.
>
> Remaining questions in my mind ...
>
> 1. Were the non-upstreamed changes to BlueZ 4 made in Tizen 2.x which
> would still have value in BlueZ 5 (and so for Tizen 3.x)?   I don't know,
> but just wanted to throw that thought into Tizen-world.   Somebody would
> need to audit that Tizen 2.x version against a vanilla BlueZ 4 and make a
> judgment call on that.
>
> 2. If BlueZ 5 is incomplete, I would assume that BlueZ 4 master is still
> being maintained?   If so, it would be nice if Tizen could upstream changes
> which weren't hacks, so that other distros still using BlueZ 4 could
> benefit.   That benefit would go beyond just mobile, right?
>
> 3. And then for Sailfish, it looks like we could just harvest Tizen
> changes into the Sailfish BlueZ 4 whether or not any upstreaming happens.
> Looks like Javier is entirely on the ball for that.   He has mentioned
> asking who is the Bluetooth "person" for Sailfish at the next community
> meeting.   So there is no further TODO for that.   It is moot until
> Sailfish gets up to Qt 5.4.
>
> Thanks, everyone!
>
>
> Cheers,
> Bob
>
>
> Cheers,
> Bob
>
>
> 7745-M: Taesoo Jun <steve.jun at samsung.com>
> 7746-M: wu zheng <wu.zheng at intel.com>
> 7747-M: Pyun DoHyun <dh79.pyun at samsung.com>
>
>
> On Tue, Aug 26, 2014 at 10:52 AM, Javier S. Pedro <jolla at javispedro.com>
> wrote:
>
>>
>> Bob Summerwill <bob at summerwill.net> escribió:
>>
>>  Hey Javier,
>>>
>>> What was the non-upstreamed functionality you saw in Tizen 2.x?
>>>
>>> I guess that it may be the case that the additions were additions to
>>> BlueZ
>>> 4 which were already present in BlueZ?
>>>
>>> Mikko - please could you confirm that Tizen 3.0 is using BlueZ 5?    Is
>>> it
>>> a full BLE stack, as far as you are aware?    Javier said ...
>>>
>>> https://twitter.com/javispedro/status/497161968508469248
>>> "Wow, Tizen folks went all the trouble to add full BLE support to Bluez4.
>>> Pairing, peripheral-role, advertising, everything."
>>>
>>
>> Hello Bob,
>>
>> that message came from the source files from the Gear2 tarball:
>> - bluez-4.101_39 contains a lot of patches (some guarded by #ifdef
>> __TIZEN_PATCH__) that enabled many LE functionality (e.g. pairing, key
>> exchanges using mgmtops, gatt server/client, advertising control, etc.).
>> Apart from this there are some Tizen specific patches that are useless
>> upstream (e.g. battery level reading).
>> - linux-3.4-exynos3250-1.0 contains the required patches to enable the
>> kernel side of this. Without them, the Bluetooth subsystem from 3.4 was way
>> too old to handle most LE operations via mgmtops.
>> They are enabled by Kconfig's CONFIG_TIZEN_WIP and CONFIG_TIZEN_RANDOM.
>>
>> Whether these are backports from Bluez5 and/or more recent kernels I
>> don't know. I suspect the Bluez patches at least are a fork, because the
>> current Bluez5 LE functionality comes, iirc, from a Google employee. So I
>> do not think they are useful for upstreaming any longer.
>>
>> My point was that many systems are stuck with Bluez4, because of the API
>> breakage that comes with Bluez5 means rewriting most of the platform BT
>> support, and because some parts of Bluez5 are not ready yet (LE among
>> others). Thus, enabling LE in Bluez4, keeping API compatibility, is very
>> interesting. An example of such platform is Jolla Sailfish, but I suspect
>> there are others (Ubuntu comes to mind).
>>
>> So while I don't think the Tizen changes are useful for upstreaming
>> (might be wrong), they would still be useful for e.g. Sailfish as a
>> temporary solution until Bluez5 support 'matures'.
>>
>> In fact, a future version of Qt will allow GATT even under Bluez4 by
>> doing something similar to what gatttool does. Qt itself does not talk with
>> Bluez5 yet.
>>
>> Javier.
>>
>>
>
>
> --
> bob at summerwill.net
>
>


-- 
bob at summerwill.net
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.sailfishos.org/pipermail/devel/attachments/20140912/db125223/attachment.html>