Hi,

On 1/8/07, arnaud guinvarch <[EMAIL PROTECTED]> wrote:
Looking at Savonet, I believe that it could be a very good tool
to build this automated brodacasting tool. As such, I would
like to know if you would know some skilled developers that could
help me in the very next future.

First, Savonet is the name of a team, I guess you're refering to our
stream generator Liquidsoap as the very "very good tool to build" your
project. I want to make this clear in case you looked too quickly at
our website and just saw our initial project (featuring many
components, from the website to an intelligent database manager). This
vision didn't last long, and we mostly only have liquidsoap left --
and a couple of small utilities. However, I do think that liquidsoap
is a good tool to build a netradio. It's generic, clean, and will
allow you to add more features easily.

The project is still alive, although moving rather slowly, in an
irregular way. The development team is quite small, compared to the
initial team -- savonet used to be a school project. I'm the only one
who knows liquidsoap's guts, and only three/four of us are still
working on the project in general (libs+tools). All of us are phd
students in computer science and none wants to give up on that, or
dedicate more time than currently to the project.

If you need somebody to help you to use liquidsoap, you don't need to
look too far: using it shouldn't be too difficult. Also, this mailing
list is there for help. My advice would be to look with somebody with
experience in functional languages (in particular OCaml) since
liquidsoap is influenced by them: the script language is functional,
statically typed with inference. Knowing OCaml can also get handy if
you need new features or fixes quickly. With more details regarding
your needs (new features or mostly support) and employment offer I
could ask a few people around me if you wish.

I take the opportunity of this mail to declare that liquidsoap is
quite stable if you avoid a few known issues:
* RTP is not stable.
* Alsa I/O depends on your hardware, AO seems to be a good replacement
for ouptut but we have none for input.
* Downsampled vorbis output eventually causes an unexplained exploding
memory consumption. We've been warned that our downsampling code sucks
so I expect to fix that when moving to the being-interfaced-with-OCaml
libresample, but I'm not sure of anything.
* There has been little testing other than on linux/x86. An endianness
bug in ocaml-lame is still open on OSX/PPC.

I hope you'll enjoy it!
--
David

Reply via email to