System Common Messages & TuneRequest not sent through MidiInterface::send method

[nativeVS]

It appears System Common Messages are not sent through the very convenient method.
After a bit of diggin I ended up adding the following lines in the hpp file which easily fixes most of that:

    else if (inType >= TimeCodeQuarterFrame && inType <= SongSelect)
    {
        sendCommon(inType, inData1);
    }
    else if (inType == TuneRequest)
    {
        sendCommon(inType);
    }

**EDIT: Ignore the following, as this turned out to be a hardware(?) problem with my MIDI interface I used for testing, see details in post further below. **

However this still didn't get the TuneRequest to be recieved in MIDI-OX.
A prjc forum post did a hypothetical fix, however the only way I seemed to actually get reliable TuneRequest output without chewing on the following SysEx bytes was to add two one empty bytes to the message (or I should say, MIDI-OX thinks now it's happy, so maybe that problem is on their end).
My quick fix was to add

mTransport.write(0x00 & 0x00);

following in the inner switch statement of the ::sendCommon method, making it:

switch (inType)
            {
            case TimeCodeQuarterFrame:
                mTransport.write(inData1);
                break;
            case SongPosition:
                mTransport.write(inData1 & 0x7f);
                mTransport.write((inData1 >> 7) & 0x7f);
                break;
            case SongSelect:
                mTransport.write(inData1 & 0x7f);
                break;
            case TuneRequest:
                mTransport.write(0x00 & 0x00);
                break;
            default:
                break;

Does anyone have any ideas why it seems to be required to make a 1 byte message into 3 2 bytes for it to be received properly? Is this a bug in MIDI-OX or some historic reasoning, or should I just look more carefeully at the MIDI spec again?

[franky47]

According to the spec, TuneRequest does not have data bytes, so there should be no need to emit them.

This sounds like a problem in MIDI-OX.

Also you don't need the mask if writing a zero, mTransport.write(0x00) is enough.

[submodify]

Send() does indeed ignore System Common Messages #369

How are you talking to Midi OX? Are you using a USB library (Teensy perhaps)? Or some external Midi Interface?

Also is it the same with other Midi monitor software like PocketMidi (with parser off), or MidiView?

[nativeVS]

Well, long story short, don't ever trust M-Audio anywhere near your MIDI (a lesson I learnt years ago and somehow still ignored). Following my musings whilst testing:

My setup for my original post that I got working was as follows:
MIDI-OX (on Windows 10) -> Midisport 2x2 -> Teensy -> Midisport 2x2 -> MIDI-OX

I generate the 1 byte message using the MIDI-OX SysEx window just typing in the hex codes; this method works happily with the other possible hex 1 byte messages.
I also just realised that I ended up sending a 2 byte message rather than erroneously stating 3 bytes above (clearly I was too tired).

My next setup was as following:
MIDI-OX (on Windows 10) -> Babyface -> Teensy -> Midisport 2x2 -> Snoize Midi Monitor (Mac)

This resulted in getting an invalid byte message in the Midi Monitor (as would be expected, I guess), however if I go back to the 1 byte write message it doesn't register the Tune Request.

Finally, I decide to bypass the Midisport as it had a history of messing up long SysEx and generally bad timing:
MIDI-OX (on Windows 10) -> Babyface -> Teensy -> Babyface -> MIDI-OX

Well, this fixed it. I guess it's time for me to badger RME again into making MIDI interfaces (or at least adding more ports), as they (along with Ploytec) seem to be the only ones to actually get it right.

[submodify]

Well, this fixed it. I guess it's time for me to badger RME again into making MIDI interfaces (or at least adding more ports), as they (along with Ploytec) seem to be the only ones to actually get it right.

I've previously tested a few commercial Midi interfaces from some reputable brands, and was surprised that all the ones I tested failed in some way (most just not conforming somehow to the Midi spec, but some would just completely crash, or go crazy), so I'm not too surprised it would be the MidiSport. Glad you found the source of that issue though.