Hi,
I'm not sure the problem comes from the track merging. Actually, I'm not
really so sure tracks are really merged, because when using fallback.skip,
the problem is not the lack of skipping. After sky(),
fallback.skip(harbor,playlist) actually skips to next track when someone
connects to harbor ! The problem is that instead of switching to the harbor
source, it stays on the playlist source. Which IMHO shouldn't happen even if
the playlist tracks are merged, the opposite (switches but doesn't skip)
would be more logical, wouldn't it ?
Anyway, implementing our own source.skipping called on_disconnect through
harbor has proved to be a working workaround.
On Mon, Jul 4, 2011 at 10:31 AM, David Baelde <david.bae...@gmail.com>wrote:
> Hi everybody,
>
> On Thu, Jun 30, 2011 at 9:09 PM, Romain Beauxis <to...@rastageeks.org>
> wrote:
> > As discussed on IRC, there may be a problem with the sky() operator,
> > which is a complex use of filters, splitting the source into 3
> > different one, applying different compressions and merging them back
> > using add().
>
> Thanks for pointing this out. I can confirm that everything goes as
> expected (from the developer viewpoint): you _can_ skip through the
> sky() operator, but the operator merges tracks (this is a known side
> effect of add()).
>
> I have tested both claims as follows:
> liquidsoap 'output.dummy(id="out",fallible=true,
> sky(playlist("~/jazz"))) add_timeout(3.,{
> ignore(server.execute("out.skip")) ; 3. })'
> liquidsoap 'clock.assign_new(sync=false,[
> output.dummy(id="out",fallible=true,
> on_track(fun(m)->log("NEW TRACK
> #{m[\"filename\"]}"),sky(playlist("~/jazz"))))])'
>
> The first tests shows successful track changes every 3 seconds. The
> second one shows only NEW TRACK for the first file, then new files are
> used but no new track is issued. Note that the second test uses the
> brand new clock API only available in HG.
>
> > As a temporary work around, I proposed to lookup if there were any
> > LADSPA plugins that would do the job.
>
> Sounds sensible. We could try to make add() respect tracks, this could
> work in the case where all sources have the same track limits (which
> is the case with three children sources which are filters of the same
> original source) but this would not be a very uniform/predicatable
> behavior.
>
> Cheers,
> --
> David
>
------------------------------------------------------------------------------
All of the data generated in your IT infrastructure is seriously valuable.
Why? It contains a definitive record of application performance, security
threats, fraudulent activity, and more. Splunk takes this data and makes
sense of it. IT sense. And common sense.
http://p.sf.net/sfu/splunk-d2d-c2
_______________________________________________
Savonet-users mailing list
Savonet-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/savonet-users
