All things related to multicast

[Zenon Mousmoulas <>]

On 23 Ιουν 2008, at 9:02 ΠΜ, Niels den Otter wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Hi,
>
> Niels den Otter wrote:
> > I'm not sure if this is 'the bug', but I got an answer from the  
> > person
> > that initiated the SAP stream yesterdam.
> >
> > He was setting up a miniSAP server (VLC) to announce a session via  
> > SDR.
> > This appeared to work too good. Problem was in the announcement
> > interval. The documentation said that the announcement interval  
> > should
> > be provided in seconds, while the sourcecode would have a
> > usleep(interval) so in microseconds.
>
> I got some more information. He was using version 0.3.4, and changed  
> the
> following in sapserver.cpp:
>
> - --- sapserver.cpp       2007-03-11 13:20:57.000000000 +0100
> +++ sapserver.cpp.patched       2008-06-23 07:40:20.000000000 +0200
> @@ -270,7 +270,7 @@
>                 /* Wait for next sending */
>                 if(should_exit)
>                     break;
> - -                nanosleep( &delay, NULL );
> +                sleep(30);
>             }
>         }
>
> This was just to get the announcement interval correct (for him).
>
> However it was still not clear to him why the original sleep was so
> short
>
> struct timespec
>              {
>                      time_t  tv_sec;         /* seconds */
>                      long    tv_nsec;        /* nanoseconds */
>              };
>
> This should indicate that the sleep was in seconds. Compilation was  
> done
> with GCC 3.3.4 under Linux 2.4.34 SMP.

You're saying the announcement interval wasn't right, but looking at  
the code I (also) can't see why that would not work.
Nor do I see a problem with the code spotted by Alan.

I'm not good at this however and since I haven't actually used  
miniSAPserver, I'm only guessing and I can't really help...

Z.