[WIP] new crone, standalone jack client

[catfact]

3rd time is the charm for PR #650

[catfact]

hrm, you tried this?

(With appropriate address ofc.)

https://github.com/monome/norns/blob/feature/new-crone/crone/osc-tests/softcut.scd

Glad the routing is basically working

[tehn]

yep. crone reports getting all of the commands, but no audio happening. will keep on it, make sure i'm not being dumb.

[catfact]

i'll be able to look on the norns tonight or tomorrow.

if you want to troubleshoot in the meantime, i guess i'd check that the softcut client is being created and connected correctly.

firstly, that there are no errors being thrown from client creation (https://github.com/monome/norns/blob/feature/new-crone/crone/src/Client.h) - these would show up amidst the early output that normally looks like:

setting up jack clients..
engine sample rate: 48000
engine sample rate: 48000
starting jack clients..
connecting ports... 
starting OSC interface...

("engine sample rate" is coming from each client init.)

and then see what jack_lsp prints once its running. this lists all the ports:

~$ jack_lsp
system:capture_1
system:capture_2
system:playback_1
system:playback_2
crone:input_0
crone:input_1
crone:input_2
crone:input_3
crone:input_4
crone:input_5
crone:output_0
crone:output_1
crone:output_2
crone:output_3
crone:output_4
crone:output_5
softcut:input_0
softcut:input_1
softcut:output_0
softcut:output_1

and with -c option also shows connections:

~$ jack_lsp -c
system:capture_1
   crone:input_0
system:capture_2
   crone:input_1
system:playback_1
   crone:output_0
system:playback_2
   crone:output_1
crone:input_0
   system:capture_1
crone:input_1
   system:capture_2
crone:input_2
   softcut:output_0
crone:input_3
   softcut:output_1
crone:input_4
crone:input_5
crone:output_0
   system:playback_1
crone:output_1
   system:playback_2
crone:output_2
   softcut:input_0
crone:output_3
   softcut:input_1
crone:output_4
crone:output_5
softcut:input_0
   crone:output_2
softcut:input_1
   crone:output_3
softcut:output_0
   crone:input_2
softcut:output_1
   crone:input_3

[catfact]

ok, also added some prints to ensure enabled and level commands are actually being handled by the softcut client

[tehn]

thanks! will test on Sunday when I am back

[catfact]

@tehn i was able to get sound and softcut working as expected,
it took longer than i expected and not even quite sure why,
but one silly mistake i made over and over was to fail to realize that each of the scripts in crone/osc-tests redefines the target address (to "localhost", so, they fail). its dumb

anyways it seems to be basically doing its thing,
usage is ~14% for 2 voices.
will try bumping up voice count to 4,6

still haven't really tested phase, vu polls on the device, or made lua attachments

[tehn]

just back now. i wasn't redefining the target address but was single-lining the osc-test scripts... will check again and let you know what i see.

14% for two voices seems pretty good? i have no actual ambition to get beyond 4 voices.

oh, likewise. i didn't test the polls yet. will do that.

[catfact]

seems ok so far with 6 voices

with rate=1 for all voices, 1-2s loops, usage is ~44%

higher rates mean more upsampling and more usage. with rate=6 (for each of 6 voices) it goes up to ~66%, and with rate=8 to ~75%. that seems problematic, though haven't actually had noticeable dropouts or other problems yet.

the other parameter, that comes to mind, to crank for worst case, is really small loop times.

gotta put this down again for now, sorry :/

[tehn]

wow. awake into softcut with the 6-voice test. initial testing is super positive. i'll continue to get into the details.

this is amazing, thank you!

[tehn]

working on oracle to connect new crone to matron.

current sc/crone deals with levels in db.amp, new-crone is float. do we have a preference? we can always do db-conversion in lua (?)

[catfact]

we can always do db-conversion in lua (?)

thats what i figured

[tehn]

could you clarify voice and channel here?

// args: channel index, voice index, level
n.sendMsg("/set/level/in_cut", 0, 0, 1.0);

there are 6 voices, right? i thought they were all mono? or perhaps i'm mis-interpreting "channel"?

[catfact]

yes, softcut voices are mono (for now)

that command is to connect an ADC input channel to a softcut voice

[tehn]

ah, understood. in is the bus which could contain adc and ext correct?

[catfact]

that's right; at the moment softcut client only has a single stereo input

(sorry yes prev answer was misleading.)

if we want, ext and ADC paths could be separate

[tehn]

EDIT: FALSE ALARM.

my sc paths and files were all messed up switching between branches. must finalize the folder re-org

---

trying to figure out if i smashed by sc config or not. when loading PolySub (via earthsea.lua) i'm seeing:

ERROR: Server-getControlBusValue only supports local servers

i'm quite possibly messing something up, but wondering if this was triggered by the new approach to supernova loading. in that case, i guess there's a way to clean up engines that require a local server?

[catfact]

wondering if this was triggered by the new approach to supernova loading. in that case, i guess there's a way to clean up engines that require a local server?

ah - yes, that is the case. getSynchronous method on busses will not work. anytime this method is used has to be refactored to use plain get, which is asynchronous with a callback. sorry...