Re: Working on keepalive messages while waiting for local connection

[g4-lisz@xxxxxxxxxxxx]

On 16.04.2018 11:27, Andreas Schneider wrote:
> On Saturday, 14 April 2018 18:35:10 CEST g4-lisz@xxxxxxxxxxxx wrote:
>> On 14.04.2018 17:58, Andreas Schneider wrote:
>>> On Saturday, 14 April 2018 15:55:20 CEST g4-lisz@xxxxxxxxxxxx wrote:
>>>> On 11.04.2018 18:09, g4-lisz@xxxxxxxxxxxx wrote:
>>>>> On 11.04.2018 17:31, Andreas Schneider wrote:
>>>>>> ould appreciate a lot if someone could show me some sample code.
>>>>>> I think that ssh_event_dopoll() will handle it and call the appropriate
>>>>>> callback. Don't use ssh_select()
>>>>>>
>>>>>> In master we have the new connector API which is easier to use, see
>>>>>> https://git.libssh.org/projects/libssh.git/tree/examples/ssh_client.c
>>>>> Thanks, I think that's the issue: The main event polling is not called
>>>>> why my code waits for a connection. But how do I get the event to which
>>>>> are attached the default handlers of libssh?
>>>>>
>>>>> I also looked at the sample code which uses the event API for polling on
>>>>> the channel. But this would not solve my issue with global keep-alive
>>>>> requests. They are sent while no channel exists. IMHO it's an issue with
>>>>> handling messages in the main loop while no channels are open.
>>>> Well, maybe I should ask it this way:
>>>>
>>>> How can I keep the internal main loop running, i.e. the default
>>>> callbacks being triggered, when nothing else is going on?
>>> You need to call of of the ssh poll functions, e.g. ssh_channel_poll() or
>>> ssh_event_dopoll() this will also trigger the standard callbacks if
>>> something is in the queue.
>> This is what I tested before (ssh_event_dopoll)... But maybe it did not
>> work because I had a "blank" event without anything attached to it. I'll
>> try it again.
>> Thanks.
> If you got it working, it would be great if you could improve the docs or 
> update the tutorial with what you've learned, this will help us and others in 
> future and is a simple way to contribute back ;-)
>
Ok. But someone should review my code first. At the moment it's more a
first try for a "proof of concept"...

Regards,
Till