Content-Type: text/plain;
        charset=utf-8
Mime-Version: 1.0 (Mac OS X Mail 12.2 \(3445.102.3\))
Subject: Re: [PATCH BlueZ v3 00/30] Major rewrite for Multi-Node and DBus
From:   Marcel Holtmann <marcel@holtmann.org>
In-Reply-To: <370824AD-3981-4FE9-9D3B-B010EC4B3EBC@intel.com>
Date:   Tue, 18 Dec 2018 18:12:59 +0100
Cc:     Szymon Janc <szymon.janc@codecoup.pl>,
        "linux-bluetooth@vger.kernel.org" <linux-bluetooth@vger.kernel.org>,
        Johan Hedberg <johan.hedberg@gmail.com>,
        "Stotland, Inga" <inga.stotland@intel.com>
Content-Transfer-Encoding: 8BIT
Message-Id: <CB93DF86-02F7-412E-BAF0-81709A258281@holtmann.org>
References: <20181215002624.9313-1-brian.gix@intel.com>
 <DEBB0CAA2616974FAE35E4B560B9A4375D8E01AF@ORSMSX101.amr.corp.intel.com>
 <17761993.WvVeqkb5i9@ix> <398288BA-8B31-4D3F-97CE-62E4CD683923@intel.com>
 <5F351471-7372-4486-B8A9-5D146AB60160@holtmann.org>
 <370824AD-3981-4FE9-9D3B-B010EC4B3EBC@intel.com>
To:     Brian Gix <brian.gix@intel.com>
Sender: linux-bluetooth-owner@vger.kernel.org
Precedence: bulk

Hi Brian,

>>>>> This patch-set has been applied upstream.
>>>> 
>>>> There is something wrong with Makefile.mesh
>>> 
>>> This was tested by us on Fedora 27 and 28, without issues.
>>> 
>>> What build platform produced these errors?
>> 
>> start adding Fedora 29 to your test platform. However this has nothing to do with Fedora 29, the whole idea of just commenting something out and committing it is just not something we do. Actually it is something that should never be done if you are operating with git. Use your personal trees for testing.
>> 
>>> Since Marcel is backing out the entire patchset, we will create a v4 patchset that removes btmesh (the non-daemon executable) entirely.  It probably won’t actually be needed in the future anyway... The DBus APIs should be sufficient.
>> 
>> Be careful to not take tools away from people that might depend on it right now.
> 
> The only functionality that was available in btmesh that was not in the meshd daemon, was the ability to request a beacon be sent.  This is done automatically by the daemon.
> 
> The balance of btmesh functionality was to be a foreground app so that all the debug output is on the console. This can be done with the meshd executable with the -nd argument (no detach).
> 
> After consulting with Inga, we think btmesh should be removed entirely.

actually the whole concept of detach is really old code from bluetoothd. Please stop using daemon() and try to fork yourself as a daemon. Leave this to system to someone else. Just start the process and enter the mainloop. The only difference should be where the log messages go. And with systemd even you do stderr, then it will be logged into journald as well. If you are in doubt, look at more recent projects like iwd on how things are done there.

Regards

Marcel