When receive too many MIDI data, it will crash

[ZZZ8]

I used rtpMIDI V1.1.11 , Esp32 DevKit v1 ,

When ESP32 receive too much MIDI data,(Receive about 10 midiNotes or midiController at the same time) 1:send 10 Notes https://i.loli.net/2019/09/11/mfhn8NFlvRYpLKP.png 2.receive 3 note on message https://i.loli.net/2019/09/11/XKuYFTlEVRe9kro.png

3. Connection failed https://i.loli.net/2019/09/11/XKuYFTlEVRe9kro.png

the esp32 will crash. What is the reason?

thx

arduino code: https://pastebin.com/z5mH5mYb

[lathoub]

Not enough information, you will have to expand. Provide code, method to reproduce, etc

[ZZZ8]

Not enough information, you will have to expand. Provide code, method to reproduce, etc

It has been added.

[ZZZ8]

single note 40notes/s Perfect single note 53notes/s Perfect single note 60notes/s Dead

[lathoub]

I have partially rewritten the library, in the 2.0.0 branch. Can you test it? (I have not tested it on a ESP32 - the endianness might not be correct, not sure)

[ZZZ8]

arduino 1.8.10 some error

error: macro "ntohl" passed 4 arguments, but takes just 1 invitation.initiatorToken = ntohl(a[0], a[1], a[2], a[3]);

error message

[lathoub]

I added more ESP32 code (especially on the endianness) and I can get the ESP32 to participate in a session :-). Hopefully you get better results now (not all MIDI messages are parsed in the AppleMIDI parser, still some way to go -looking for volunteers!)

[designerfuzzi]

a NoteOff is actually a NoteOn Message with Velocity lower then 64 or more often (usually) with velocity (zero) 0. Meaning the Callbacks for both can be handled at once and your amazing lib could follow the standard and handle NoteOff as NoteOn with 0 velocity. This way you can also handle stuck notes that have been never closed and implementation of active sensing makes sense then.

Why? Because as it is now the lib or programmer using it have to compare NoteOn and NoteOff Messages to know which notes are open and largely depend on AllNoteOff messages to close all open while AllNoteOff messages are not send by all AppleMidiNetwork emitters.

Hint: solution maybe instead - receiving Opening Note Messages on already open Notes would re-trigger the corresponding callback while NoteOff (aka velocity 0 NoteOn) would close the same note-pitch instead of triggering a separate callback which does know nothing about the open notes.

[lathoub]

Branch 2.0.0 (heavily modified the parser for combined messages, incl large SysEx)) has been stress tested (Active Sensing at 270ms, Timing Clock at 250 BPM, large repeated Sysex and continuous NoteOn/Off). Can you please test branch 2.0.0 @ZZZ8 ?

[ZZZ8]

OK . tomorrow ,i test it.

[lathoub]

Thanks - make sure you test the latest 2.0.0 branch

[ZZZ8]

applemidiTest.xlsx

@lathoub Test results in the Excel document

Thank you for your support

[ZZZ8]

More stable than last time , No crashes when I send about 20 Notes.

[lathoub]

Thanks @ZZZ8 Interesting results!

Do I understand it correctly that notes are dropped? (miss). Missing notes without seeing errors in the log is strange - a note in rtpMIDI is surrounded by protocol bytes and counters. So missing a note without causing other protocol error is strange.

Are you able to include Wireshark in your tests?

[lathoub]

@ZZZ8 Are you sure you are using the latest branch v.2.0.0?

I installed Reaper v6.03 earlier and that revealed some Journal section errors in the code - fixed and uploaded. (Journal section was much larger than what I have seen in the past)

[ZZZ8]

666, You modify very fast .Thanks .

I haven’t had time to test the latest version,today. I will test it,these two days

发送自 Windows 10 版邮件https://go.microsoft.com/fwlink/?LinkId=550986应用

发件人: lathoubmailto:notifications@github.com 发送时间: 2020年2月15日 19:32 收件人: lathoub/Arduino-AppleMIDI-Librarymailto:Arduino-AppleMIDI-Library@noreply.github.com 抄送: ZZZ8mailto:ZLC@MSN.CN; Mentionmailto:mention@noreply.github.com 主题: Re: [lathoub/Arduino-AppleMIDI-Library] When receive too many MIDI data, it will crash (#71)

@ZZZ8https://github.com/ZZZ8 Are you sure you are using the latest branch v.2.0.0?

I installed Reaper v6.03 earlier and that revealed some Journal section errors in the code - fixed and uploaded. (Journal section was much larger than what I have seen in the past)

― You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/lathoub/Arduino-AppleMIDI-Library/issues/71?email_source=notifications&email_token=AJYDUI2RUQQGOROBTB7U7KLRC7HDPA5CNFSM4IUTXXVKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEL3IJBA#issuecomment-586581124, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AJYDUI2VVA4W4BMZ7Q26ULLRC7HDPANCNFSM4IUTXXVA.

[lathoub]

Let me know how the tests go - I tried with REAPER as well (v6.03/64) on MacOS with no drops (branch 2.0.2 rc1)

[ZZZ8]

[cid:image001.png@01D5ECC9.863E2920] 1\ Input notes in the DAW,8notes,9notes,10notes. and play it . 2\ look at the arduino Serialprot , Number of note.On

发送自 Windows 10 版邮件https://go.microsoft.com/fwlink/?LinkId=550986应用

---

发件人: lathoub notifications@github.com 发送时间: Monday, February 24, 2020 12:15:28 AM 收件人: lathoub/Arduino-AppleMIDI-Library Arduino-AppleMIDI-Library@noreply.github.com 抄送: ZZZ8 ZLC@MSN.CN; Mention mention@noreply.github.com 主题: Re: [lathoub/Arduino-AppleMIDI-Library] When receive too many MIDI data, it will crash (#71)

Let me know how the tests go - I tried with REAPER as well (v6.03/64) on MacOS with no drops (branch 2.0.2 rc1https://github.com/lathoub/Arduino-AppleMIDI-Library/tree/v2.0.2-rc.1)

― You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/lathoub/Arduino-AppleMIDI-Library/issues/71?email_source=notifications&email_token=AJYDUI6TJOEYSMENUXYC35LREKOKBA5CNFSM4IUTXXVKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEMV73PI#issuecomment-590085565, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AJYDUI3UVXSLOW5WBHWZQ3LREKOKBANCNFSM4IUTXXVA.

[lathoub]

added Session Initiator in branch 2.0.2 (and example)

[lathoub]

more Initiator testing (see Initiator example), can you test the latest branch feat/2.0.0 (I have not made a release - fetch the latest from feat/2.0.0) @ZZZ8

[lathoub]

Fixed in the latest release 2.0.3 (now in the master branch)