Nevermind, I spotted the idiocy I was doing, trying to use a system
error type for a custom error! The solution is to declare an error
constant and display a error description.

Edward

On 27/08/2015, Edward Bartolo <edbarx@???> wrote:
> Oooops, forgot the attachment! Here it is.
>
> On 27/08/2015, Edward Bartolo <edbarx@???> wrote:
>> I am trying to write a C function to start the wired network
>> interfaces eth0, eth1 up to eth2. This is how I wrote it, but I would
>> like your opinion about it especially where I should declare err.
>>
>> The attachment is best viewed with medit with a tab space of 2.
>>
>>
>> On 27/08/2015, Edward Bartolo <edbarx@???> wrote:
>>> I don't think it is necessary to daemonise the backend as connection
>>> to and disconnecting from a network is not something that is done
>>> continuously. Therefore, in my humble opinion, running the backend on
>>> request seems to be the best approach. This also avoids additional
>>> complexity of requiring a dedicated init script. An SUID belonging to
>>> root is enough for the backend to be allowed privileges automatically
>>> whenever it is invoked. This means, the post installation script will
>>> only need to "chmod u+s backend" and create a launcher on user
>>> request.
>>>
>>> I will now test an installation to /usr/bin of both backend and
>>> frontend using an SUID for the backend. If everthing goes well, it
>>> would mean, the time for an ALPHA release of a .deb package for
>>> netman, is possible now.
>>>
>>> I included more functionality in the frontend to recognise the
>>> existence of an /etc/network/interfaces displaying and option to
>>> connect to eth0 if that is found. I also included a compiler
>>> conditional directive to compile the frontend so as to bypass the
>>> requirement of sudo.
>>>
>>> When I am ready, and I think the project can be package, I will upload
>>> to git.devuan.org
>>>
>>>
>>> Edward
>>>
>>> On 26/08/2015, Isaac Dunham <ibid.ag@???> wrote:
>>>> On Wed, Aug 26, 2015 at 02:27:57PM +0200, tilt! wrote:
>>>>> On 08/26/2015 01:36 PM, Irrwahn wrote:
>>>>> >[...]
>>>>> >Or, even better, you could easily pass the IF name as an additional
>>>>> >parameter to the backend (and possily even use it as additional
>>>>> > component
>>>>> >to construct the interface file names).
>>>>>
>>>>> An easy way to obtain a list of interface names on Linux is:
>>>>>
>>>>> awk 'NR>2{gsub(/:/,"");print $1}' /proc/net/dev
>>>>
>>>> Or "ls /sys/class/net".
>>>> For listing non-loopback devices only, use
>>>> ls -d /sys/class/net/*/device | cut -d/ -f5
>>>>
>>>> For wireless, check for the files "phy80211" or "wireless".
>>>>
>>>> The equivalent can be done trivially in C with readdir(); ask if you'd
>>>> like
>>>> an example.
>>>>
>>>> HTH,
>>>> Isaac Dunham
>>>> _______________________________________________
>>>> Dng mailing list
>>>> Dng@???
>>>> https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng
>>>>
>>>
>>
>