Hi !
Le mardi 23 juin 2009 21:57:44, Robert McAuley a écrit :
> Could it be the smooth_add, even though the delay is only 0.5?
I don't think so.
However, you don't need such a short length for the request.dynamic.
For my tests, I use the default parameters, which is 10 seconds, and
it seems to work fine.
Also, but I don't think this is your case, if you use an operator that buffers
data in advance, such as the crossfades, you should have enough
remaining data so that liquidsoap has enough time to prepare the next
song when the remaining data of the ending source is being buffered.
> I would have thought 2s would be enough to get Liq to buffer what it
> needs. (or perhaps my conservative parameter is in the wrong place?)
The conservative parameter seems correct. The new behaviour takes into
account the *estimated* remaining time. From times to times, it is not
impossible
that it takes two songs if the remaining time is not correctly guessed, in
particular when starting.
However, my tests locally showed a correct behaviour with the default
length and only conservative=false..
8<---------------------------------------------------------->8
# Log to stdout
set("log.file",false)
set("log.stdout",true)
set("log.level",5)
def f () =
file = list.hd(get_process_lines("find ~/documents/zic | grep \".mp3$\" | sort
-R 2> /dev/null | head -n 1 2> /dev/null"))
print("\n\n\nGot file: #{file}\n\n\n")
request.create(file)
end
s = request.dynamic(f,conservative=false)
s = mksafe(smart_crossfade(s))
output.pulseaudio(s)
8<---------------------------------------------------------->8
Romain
------------------------------------------------------------------------------
_______________________________________________
Savonet-users mailing list
Savonet-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/savonet-users
