[Icecast] Open source streaming project in need of developers
Geoff at QuiteLikely.com
Thu Dec 20 12:50:24 PST 2007
David Baelde wrote:
> You didn't describe much of your bugs or the specific features of your
> playout system...
hmmm, I thought I did describe the playout system, but I'm happy to
elaborate if you want more details.
But you're right about the bugs. The critical (i.e. urgent) ones in no
1. PRS will disconnect from Icecast after a certain amount of time (this
varies). When it does, it will then repeatedly connect and disconnect
until it eventually dies. I'm thinking something else is going wrong
and it's spiralling out of control.
2. URL events (i.e. relaying from a live stream) don't work. This has
always been rather tenuous, but now it's fully broken.
3. I've not looked hard at it, but on the face of it at least, Shoutcast
streaming seems to be broken.
4. Due to the above Shoutcast problem and the fact that I want to run it
in Icecast mode anyway, I wrote a patch (my only ever substantial C
work) to implement title streaming for Icecast 2 servers. You can see
It worked fine with stock Icecast 2.3.1 but when I upgraded to recent
Icecast (revision 1441 I think), PRS would stop playing audio shortly into
a track and stay silent until the next one. I've not had my patch reviewed
by a competent programmer, but I rather suspect my patch or something else
in PRS rather than Icecast.
These are the critical bugs that I know of. Fixing these may reveal others
not currently testable.
> It seems to me that it'd be best to focus new
> efforts on the automated scheduler pickings songs from the database.
> In that area there is no really good open-source solution as far as I
> know. There are only several specialized solutions.
This project is not new - we've been using it for 5 years. It's simply
in need of new maintainers.
> Playout is quite independent from scheduling: streamers can interface
> easily with any scheduling program.
There may well be some sense in the future in separating the playout from
the scheduling in PRS, but right now they're intertwined.
> In particular our baby (liquidsoap) does crossfading, dynamic compression
> and (much) more.
I took a brief look at it when considering whether to move to something
else rather than stick with what we have. But it appeared to me that you
need to learn some kind of programming language type thing in order to tell
it what to do.
The ACB Radio project requires software that can be operated for the most
part by reasonably non-technical types who are more worried about which
program should air in the next hour than finding the right process ID to
send a HUP to. Most have never programmed anything or ever used a CLI.
> I don't mean to advertise too much our tool. It's just a hobby for all
> of us, we don't get money for it and you won't get paid support for
> it. I just believe that it could be a loss to spend energies in
> developing other complex streamers without a good reason and a serious
> reviewing of existing tools, not only liquidsoap -- I know that some
> people have some dirty-patched versions of ices that do crossfading,
> for example.
And I don't mean to put your project down, it sounds well worth looking at.
but it doesn't seem to fit our requirements, and since our code already
does (when it's working properly) most of what we need it to do, it would
make sense to me to stick with it.
> I don't offer any help as I'm more than busy with my own project and
> life, but I'd be happy to follow any discussion on the design of an
> usable and flexible radio scheduling software.
Well I'm happy to explain how PRS does its scheduling (as much as I
understand it) and you're of course able to pinch code from the project (in
accordance with the license).
BTW: If anyone wants to hear PRS in action (the old codebase on the old
server), check out either http://acbradio.org:8000/mainstream or
http://acbradio.org:8000/cafe (/cafe-low for modem users).
> Good luck!
And you. I'll take a closer look at your project some time when I can put
a few hours into it.
More information about the Icecast