Norns: alternative grids

Helen · 2021-01-27T19:16:24.608Z

Direct to USB, no hub.

I’ve switched around among the USBs on each shutdown/restart - there have been a few!) so have had it plugged in on (I think) 3 different ones so far.

Edited to add: I’m looking at the core.lua script (midigrid/core.lua at master · miker2049/midigrid · GitHub) and although I have absolutely no idea what’s going on, I’m vaguely wondering if there’s something in the supported devices section needs looking at more closely:

function midigrid.init()
    local supported_devices = {
        apcmini = 'apc mini',
        launchpadmk2 = 'launchpad mk2',
        launchpadpro = 'launchpad pro 2',
        launchpad = 'launchpad',
        launchpadmini = 'launchpad mini'

Does it, do we think, need something like this adding?:

launchpadmini3 = 'launchpad mini mk3'

Anyway, that’s enough for me for one day - 8pm here in London after my usual 5.30am start - and I think my brain’s about to starting leaking out of my ears…
Thanks everyone who’s chimed in so far, it’s so great to know there are others who are interested in the problem-solving side of it all.

JaggedNZ · 2021-01-27T20:06:50.886Z

@ngwese any chance you can explain what might trigger the “midi inconsistency for device” ? My limited low level C skills fail me. Is this triggered by a flood of midi events or a flood of midi connect events?

@helen you might want to try a powered usb hub if you have one.

I’m not sure I’ve stated this here but I strongly recommend a powered usb hub if you are using more than one device as lighting all the LEDs will cause low voltage drop outs on your pi/Norns

Helen · 2021-01-27T20:17:27.474Z

I’m afraid I don’t have a powered USB hub - I’ll have a look online and see what’s available (is there any minimum spec that would be useful?)

The Launchpad is the only other device I’m planning on using with the Shield - audio input would be most likely from whatever synth’s headphones or line out direct to Shield’s audio in so wouldn’t be expecting the Shield to power it.

zebra · 2021-01-27T20:21:40.445Z

JaggedNZ:

any chance you can explain what might trigger the “midi inconsistency for device”

pretty weird one…

the only thing that pops out to me is that this loop index is a uint8_t, and it is compared with a ssize_t. if there are >256 bytes in the buffer then that could get borked in the loop comparison and/or the result comparison. (why would there be such a huge buffer in one read cycle? idk. systex?)

might be worth changing uint8_t i; to ssize_t i; here:

github.com

monome/norns/blob/main/matron/src/device/device_midi.c#L245-L249


      
              state->msg_pos = 0;
          }
          
          static inline void midi_input_msg_acc(midi_input_state_t *state, uint8_t byte) {
              if (!state->msg_started) {

ngwese · 2021-01-27T20:48:49.201Z

I can see one or two things which should get buttoned up in the code (such as also bounds checking the against the buffer the individual bytes are being written into, currently only 128 bytes). Regardless during development I was never able to generate buffer sizes even remotely close to 128 bytes in size even while hammering norns with sysex data dumps from other devices USB devices.

The error message was intended to indicate that some multi message MIDI data (such as running status or sysex message) was interleaved with a MIDI data that is not allowed during those multi message transmissions. Another possibility would be a device sending running status or sysex then disconnecting in the middle. The current logic will completely toss the contents of the ALSA/kernel level MIDI receive buffer if the contents seem to be corrupted in an effort to get back to a known good state. Of course this assumes that the norns level code for handling incoming MIDI is bug free…

I can take a closer look this evening when I have more time (unfortunately I don’t have access to one of the targeted midi grid devices so I’ll be flying blind to some degree).

UPDATE: one additional thought is that maybe the device is sending “active sense” MIDI messages and those are not getting handled correctly. Will have to find something which emits active sense here.

JaggedNZ:

as lighting all the LEDs will cause low voltage drop outs on your pi/Norns

…if this is something which is happening then it is within the realm of possibility that this could trigger the inconsistent midi data errors.

DoS · 2021-01-27T21:03:53.679Z

Helen:

Does it, do we think, need something like this adding?:

launchpadmini3 = 'launchpad mini mk3'

Looking in /we/dust/code/midigrid/lib/devices/supported_devices.lua, this is what I see there:

local supported_devices = {
midi_devices = {
{ midi_base_name= ‘launchpad’, device_type=‘launchpad’ },
{ midi_base_name= ‘launchpad mini’, device_type=‘launchpad’ },
{ midi_base_name= ‘launchpad mini mk3 2’, device_type=‘launchpad_minimk3’ },
{ midi_base_name= ‘launchpad mini mk3 2 2’, device_type=‘launchpad_minimk3_128’ },
{ midi_base_name= ‘apc mini’, device_type=‘apc_mini’ },
{ midi_base_name= ‘block 1’, device_type=‘livid_block’ },
–[[ Known devices to be implemented, help required!
Old launchpad mini’s have user set hardware ID 1 - 16:
e.g. ID 4 appears as midi_device.name “Launchpad Mini 4”
{ midi_base_name= 'launchpad mk2 ? ', device_type=‘launchpadmk2’ },
{ midi_base_name= ‘launchpad pro 2’, device_type=‘launchpadpro2’ },
{ midi_base_name: ‘launchpad mk3’ device_type:‘launchpadmk3’ },
]]–
}
}

The issue here might be that you need to install the repo at https://github.com/jaggednz/midigrid, which is the fork that @JaggedNZ has got working - I was never able to get the original miker2049 script working with the Launchpad Mini Mk3.

Helen:

Edited to add: I’m looking at the core.lua script (midigrid/core.lua at master · miker2049/midigrid · GitHub )

okyeron · 2021-01-27T21:31:46.977Z

ngwese:

UPDATE: one additional thought is that maybe the device is sending “active sense” MIDI messages and those are not getting handled correctly. Will have to find something which emits active sense here.

active sensing should be ignored by norns midi.lua.

I have keyboard that sends active sensing if something needs testing/confirmation

JaggedNZ · 2021-01-27T22:22:28.466Z

Your right, looks like there is some missing code from my branch to support mk2 and MK3 devices. I’ve found the code on someone else’s fork, I will try merge it later today.

I think the midi inconsistency error is a red herring in terms of getting the device in question working. From memory the later launchpad devices present multiple midi interfaces some are used for standard midi and some for other things (ableton live?) I suspect the error is triggered by one of these other interfaces.

DoS · 2021-01-27T22:47:04.051Z

JaggedNZ:

the later launchpad devices present multiple midi interfaces some are used for standard midi and some for other things (ableton live?)

Yes, this is exactly right - hence this in /lib/devices/supported_devices.lua:

DoS:

{ midi_base_name= ‘launchpad mini mk3 2’, device_type=‘launchpad_minimk3’ },
{ midi_base_name= ‘launchpad mini mk3 2 2’, device_type=‘launchpad_minimk3_128’ },

But maybe I added that (edit for clarity: to my local copy of supported_devices.lua) myself? It’s been a while since I got this running and I’m now confused.

Helen · 2021-01-28T05:31:28.616Z

Wow, a massive thank you to @zebra, @ngwese, @DoS, @okyeron and @JaggedNZ for taking the time to look into this and batting ideas around. Most it goes waaay over my head but I think it’s still safe to say you folks are awesome!

For me, now, the things I need to do are: uninstall whichever repo I installed at the start then reinstall @JaggedNZ’s later (after the code update/merge) - and get hold of a powered USB hub in the meantime.

Thanks again for all the care and attention you’ve all put into this, it really means a lot.

Update: I’ve ordered this powered USB hub:

https://www.amazon.co.uk/Powered-RSHTECH-Charging-Portable-Splitter-Red/dp/B07RK3S4KF/ref=sr_1_3

…and this (not cheap) USB-B to USB-B cable to connect Shield and hub, because (a) I don’t have any B-B cables, got almost every other flavour, and (b) it does fast data transfer:

https://www.amazon.co.uk/C2G-Metre-Cable-Speed-Transfer-Black/dp/B00436GQN2/ref=sr_1_53

Both should be here tomorrow (Friday) so here’s hoping they do the trick…

delineator · 2021-01-28T08:21:17.886Z

You’re correct that LP mini mk3’s present multiple midi devices over USB.

I got two LP mini mk3’s working with code changes you can see in this PR on to the miker2049/midigrid repo:

github.com/miker2049/midigrid

Add initial launchpad mini support

miker2049:vgrid_wip_apc ← robmckinnon:vgrid_wip_apc

opened 10:19PM - 23 Sep 20 UTC

robmckinnon

+98 -15

That PR also has a LP mini mk3 LED colour palette defined to emulate the vari-bright grid brightness levels.

Should I make the same PR to a different repo?

JaggedNZ · 2021-01-28T09:07:47.154Z

I would appreciate a PR to jaggednz/midigrid. I just managed to trash my local master attempting to merge across the changes.

UPDATE: I’ve managed to recover and it looks like I’ve already done a selective merge in the past. Just need to tidy-up and it should be in good shape.

DoS · 2021-01-28T09:42:26.210Z

delineator:

That PR also has a LP mini mk3 LED colour palette defined to emulate the vari-bright grid brightness levels.

20 characters of excellent news!

JaggedNZ · 2021-01-28T10:00:21.374Z

@DoS I suspect you are on the vgrid_wip branch of my fork of midigrid, if you are able to please move to my master branch and confirm it works?

@Helen and anyone else who has used ;install with my fork, please try to update midigrid from project manager in maiden. With any luck it should work with the MK3 now.

Sorry to everyone impacted by this, looks like my local copy was a mess and I did not notice it was out of sync with what was on my github fork. I’m looking at how we can best manage this project going forward, I’m open to suggestions here.

If you have any issues it would be helpful to capture the device loading section in maiden, it will look something like this, but vary based on your device(s):

maiden output

including /home/we/dust/code/midigrid/lib/supported_devices.lua
Default 64 layout init
vgrid init with layout: 64
core midi devices
1 table: 0xbd4870
2 table: 0xbc08e0
finding device: 1 with name virtual
finding device: 2 with name Launchpad Mini
Dev2 FD launchpad
mounted_devices
2 launchpad
Loading midi device type:launchpad on midi port 2
including /home/we/dust/code/midigrid/lib/devices/launchpad.lua
including /home/we/dust/code/midigrid/lib/devices/generic_device.lua
Connected devices:
2 table: 0xc4ac58
attaching devices:
2 table: 0xc4ac58
device._reset

Helen · 2021-01-28T10:31:40.299Z

Thank you! Will try that a bit later

MentalSandal · 2021-01-28T14:45:45.388Z

A few days ago I had a quick go at the older version of Midigrid, but encountered more or less the same issues @Helen reported. I didn’t have much time to look at the code, but suspected something was not set properly in supported_devices.lua

I’m really happy to report that @JaggedNZ 's latest commit works perfectly!

I managed to use my Launchpad Pro MK3 by adding this line in supported_devices.lua

{ midi_base_name= 'launchpad pro mk3', device_type='launchpad_promk3' },

And I also had to duplicate and rename launchpad_minimk3.lua to launchpad_promk3.lua

Some further changes might be needed, but that was a quick way to start playing during my lunch break.

A massive thanks to everyone involved in solving this issue, and providing launchpad support in the first place!

Helen · 2021-01-28T15:00:07.463Z

That’s so good to hear! I finish work for the day soon and will try it later!

Did you update the repo as @JaggedNZ suggests upthread (3 posts back)?

MentalSandal · 2021-01-28T15:33:24.257Z

I briefly had miker2049’s repo installed to see if I had the same error, so I had to delete it and install JaggedNZ’s fork again. In your case, updating from Maiden should work just fine. Enjoy!

Helen · 2021-01-28T16:32:30.800Z

WOOHOO! happy dance

Delighted to report that I’ve managed to get the Launchpad Mini Mk3 working with grid-test - I’ve put a short vid up on my Insta, running through each of the four modes (fade, chase, diagonal, random).

https://www.instagram.com/p/CKmESPThJeg/

As before, huge thanks to @zebra, @ngwese, @DoS, @okyeron and @JaggedNZ for all the help, support and figuring out the difficult stuff!

Next up: Can Helen break awake? Watch this space…

DoS · 2021-01-28T16:34:54.075Z

Helen:

Can Helen break awake? Watch this space…

Good luck with that! Glad you got it working. Now for the fun to start (and a second Launchpad Mini Mk 3 to become necessary - it only took me a week or two of fun and games to realise that 128 is so much better than 64)