Assortment of recent bugs

[yatli]

• [x] Chain mode isn't effective
• [x] LEV of all tracks stuck at 127 on chain write
• [x] Chain write do not reliably activate all selected tracks
• [x] Chain write on a tight time budget result in off-beat tracks. The problem goes away if I stop/restart playback.
• [x] Save results in wrong kit name
• In X.04 with trig chain of >=3 tracks, sometimes one of the tracks is double-triggered (b-boom)

[yatli]

C:\Users\Yatao\Documents\Arduino\hardware\MIDICtrl20_MegaCommand\avr\cores\megacommand\MD\MDMessages.cpp:214:57: warning: left shift count >= width of type [-Wshift-count-overflow]
 uint8_t MDMachine::get_model() { return model &= ~(1 << 17); }

[jmamma]

 uint8_t MDMachine::get_model() { return model &= ~(1 << 17); }

Bit 17 is set as tonality. So we must clear it when returning model type ??

[jmamma]

Chain write on a tight time budget result in off-beat tracks. The problem goes away if I stop/restart playback.

Yeah this one is on the list.

[jmamma]

LEV of all tracks stuck at 127 on chain write

This is expected behaviour. CONFIG -> MACHINEDRUM -> NORMALIZE. (LEV is boosted to 127, and VOL is lowered to compensate.)

[yatli]

uint8_t MDMachine::get_model() { return model &= ~(1 << 17); }

Bit 17 is set as tonality. So we must clear it when returning model type ??

uint8_t return type will automatically clear that

nvm, this should be fine.

[yatli]

This is expected behaviour. CONFIG -> MACHINEDRUM -> NORMALIZE.

I see. I think I hear the difference because I have an LFO on VOL -- should turn normalization off.

[jmamma]

uint8_t return type will automatically clear that

good point.

[jmamma]

I see. I think I hear the difference because I have an LFO on VOL -- should turn normalization off.

The is normalisation code should scale LFOs accordingly, but i think the relationship is not linear so there may be some differences.

11 void MDMachine::scale_vol(float scale) {
 12   DEBUG_PRINT_FN();
 13   params[MODEL_VOL] = (uint8_t)((float)params[MODEL_VOL] * scale);
 14   if (params[MODEL_VOL] > 127) {
 15     params[MODEL_VOL] = 127;
 16   }
 17   if ((lfo.destinationParam == MODEL_VOL) && (lfo.destinationTrack == track)) {
 18     params[MODEL_LFOD] = (uint8_t)((float)params[MODEL_LFOD] * scale);
 19 
 20     if (params[MODEL_LFOD] > 127) {
 21       params[MODEL_LFOD] = 127;
 22     }
 23     lfo.depth = params[MODEL_LFOD];
 24   }
 25 }
 26 
 27 float MDMachine::normalize_level() {
 28   DEBUG_PRINT_FN();
 29   if (level == 127) {
 30     return 1.0;
 31   }
 32 
 33   float scale = (float)level / (float)127;
 34   level = 127;
 35 
 36   scale_vol(scale);
 37   return scale;
 38 }

[jmamma]

Seems like the timing issue only occurs when MIDI tracks on grid 2 are loaded.

[jmamma]

3ad60ec

Timing feels good now. Can you confirm @yatli

[jmamma]

Save results in wrong kit name
In X.04 with trig chain of >=3 tracks, sometimes one of the tracks is double-triggered (b-boom)

I can't replicate these.

I also cant replicate your merge bug with GND-X machines.

[yatli]

I can't repro the trig chain problem now.

[yatli]

Another bug -- chaining while playing result in wrong automation -- I have a track without slides, but chaining to it result in a knob being modulated from 0 to 127 and back again rapidly.

[jmamma]

LFOTrack ?

[yatli]

oh didn't thought of that. But when I stop/start it's gone

[jmamma]

Can you replicate the chain movement that got you in to the state?

[jmamma]

@yatli

/Applications/Arduino_nightly.app/Contents/Java/hardware/MIDICtrl20_MegaCommand/avr/cores/megacommand/CommonTools/RingBuffer.h:111:16: error: invalid conversion from 'volatile void*' to 'void*' [-fpermissive]
     memcpy(buf + wr, src, s * sizeof(C));
            ~~~~^~~~

?

[yatli]

Add -fpermissive to g++ flags

[yatli]

FYI current arduino platform.txt options:

compiler.cpp.flags=-std=gnu++1z -c -Os {compiler.warning_flags} -MMD -Wno-attributes -Wno-write-strings -fpermissive -fno-exceptions -fno-threadsafe-statics -fno-rtti -ffunction-sections -fdata-sections -flto=jobserver -Wl,--gc-section,--defsym=__stack=0x8021ff,--section-start,.data=0x802100,--defsym=__heap_end=0x80ffff,--relax -I{build.core.path}/Midi -I{build.core.path}/GUI -I{build.core.path}/CommonTools -I{build.core.path}/Elektron -I{build.core.path}/MD -I{build.core.path}/MidiTools -I{build.core.path}/Sequencer -I{build.core.path}/MNM -I{build.core.path}/PageTools -I{build.core.path}/MDSketches -I{build.core.path}/MDPages -I{build.core.path}/A4 -I{build.core.path}/SPI -I{build.core.path}/Wire -I{build.core.path}/Adafruit-GFX-Library -I{build.core.path}/Adafruit_SSD1305_Library -I{build.core.path}/SdFat -I{build.core.path}/SDCard -I{build.core.path}/MCL

[jmamma]

nice

[yatli]

Converted the X tunnel song to MCL -- timing is good. Also tight chain writes are robust! 🎖

[yatli]

Trying to extract that project for you via MegaCom:

[yatli]

MegaCom behavior is interesting -- matches the design goals -- more bandwidth on simpler pages (17KB/s on MIXER), and "nicer" on fancy pages (7KB/s on GRID)

Screensaver = 30KB/s

[yatli]

X_mcl.zip

[yatli]

hmm I still cannot activate chain mode

[jmamma]

Haven't even had a chance to look at megacom yet. you'll need to recreate the binaries because last time I tried they didnt work.

I'll be heading off for a few hours. back later tonight.

[jmamma]

hmm I still cannot activate chain mode

what do you mean?

[yatli]

Set row 0 to chain to row 1, manual Chain write row 0, expecting it to jump to row 1 after playback Not jumping, not displaying the estimated jump time below current time

[jmamma]

there used to be a chain bug on row 0. maybe it came back.

[yatli]

Tried row 1 no banana

[jmamma]

Did you set chain mode to AUTO

[jmamma]

Loops need to be greater than 0 per slot

[jmamma]

@yatli can you double check and report back.

[jmamma]

⏳

[yatli]

Woke up

[jmamma]

Wake up Samurai, we've got a firmware to burn.

[yatli]

Just tested -- chain update on any row, loop=1, row=next, cannot activate. Side effect is, kit name is lost after the update

[jmamma]

Chain mode ?

[yatli]

Yes, the button 3 menu, use encoder3 to select all tracks, set row to next, set loop to 1, release button 3 -- kit name is now blank.

[jmamma]

No i mean... from grid page.. hold shift 2. Set CHAIN mode from MANUAL to AUTO

[yatli]

It works! AUTO never worked for me before -- I was sticking with MANUAL all along

[jmamma]

haha 😆

[jmamma]

I think MCL 3.10 is done. I've updated the changelog.

[jmamma]

That means it's time.

[yatli]

[yatli]

Yahoooo

[jmamma]

preparing binaries...

[jmamma]

Ready to go here. I sent you a message elsewhere on github for your verification .

[jmamma]

okay tested here.

🚀

[jmamma]

Going live...