Hi,

Just a few quick notes..

 Ah ha! Implementation, It's nothing that I've spent a great deal of time on
since I'm more concerned at the moment of getting some good gui components
together.

I will insist on the implementation and I believe it is important to
keep that in mind a bit. Liquidsoap has a very specific design which
allowed for great flexibility for the user and great modularity for
the devs. However, this has a price: the power of our scripts means
that their behaviour is not easy to predict without running it -- that
may remind you of the halting problem.

ideal feature would be able to pipe out the output stream to the DJ's desk,
so they know whats going out.

If the output is streaming to icecast, then is feasible to get the
server/mount information for it and stream the data from icecast. But
there will be a lag. A lower latency solution could be possible, but
only for low-latency outputs (our experimental rtp or its successor)
which have explicitely been added in the liquidsoap script. If the
node is not in the script, the feature won't be there. Also, it is not
trivial to know that two outputs are actually streaming the same
content.

 Amber - Available but Blank

Currently, you can know whether or not a stream is blank, but not from
the output nodes. These don't compute this data, and won't know it if
it's computed by one of our *blank() nodes. Of course we can perform
changes (like adding a server command to some of the *blank()
operators so that they advertise the current blank/not-blank status of
the stream) but keep in mind that some things will remain difficult in
our model (even if you have blank computation nodes, it won't be
trivial to know that they work on the stream which is output by a
given output node).

 The other main part of the status is 'What's next?', this can then let

I'm worried by what you mean here: do you intend to display
information about the next song to be played ? It is currently
difficult to know without inspecting the script: you have to examine
which nodes are on the road from the output to the sources, and the
switches on this road might have a behaviour which is difficult to
guess in general. However, in many cases, you know that what's next
should be the next file in a playlist, which you can get from
liquidsoap. But if there is a jingle insertion based on random(), it
becomes difficult to predict when it will be triggered.

I hope that I didn't discourage you with all these remarks. I just
wanted to make sure that you know enough of the tool that you are
interfacing with.

Cheers!
--
David

Répondre à