So commands might look like

Bend 250

Note On A

Bend 0

Note On B

Since both bend commands occur on the same tick, will both bends be applied correctly, or will both notes get Bend 250, or will both get Bend 0?

I can't seem to find anything in the specs that specifically mention it.  I was under the assumption that only the last bend would be used, since I have been working with files that typically are single-note-at-a-time.  However,  now that I am looking at some multi-note files, I am wondering if I was mistaken.  If both bends DO apply, then I will need to modify my duplicate-command pruning function to be more selective.

That made it much harder than it should be to update things in the editor and keep track of other things like file location indexes and such.  I finally got fed up enough with it a few months ago to re-write the core from scratch.  The new code is so much more efficient, organized, and flexible.

As I'm running all of my old test files through the new editor, I'm occasionally finding things that either weren't a problem in my old code, or were and I couldn't fix, or were things that I never realized were a problem.

The most recent that I just discovered is that, unbeknownst to me, running status is NOT ended if you hit a META event.  It just picks right back up after the META event!  Took me a bit to figure out what was happening, as I have always assumed that META events could not be running (they can't) and as such, the running status flag should be disabled.  So I disabled it and always wrote the command byte for the next event.  But a test file I have has only a single Note-On command in the whole file and all the rest are running status, with META text events between groups of notes.  My saved copy of this file should have been near-identical, except now it had a bunch of Note-Ons that weren't present in the original.  Now that I've realized how running status ACTUALLY behaves around META events, I am able to reproduce an exact copy of the test file after loading, parsing, sorting events, and then rebuilding at the end.

Just thought it was interesting and a good bit of information to know for those who are working on their own software.

The new version of my editor will be posted on my Google Drive when it is "ready", and then I can finally finish up the next soundtrack-in-progress for the SNES MIDI Remaster Project.  If you are more interested about the inner workings of the editor, you can check some of my older posts on this forum as well as posts over on the Gamemaker forums.

https://forum.gamemaker.io/index.php

More to come.

Taking debouncing aside and velocity, a diode matrix scan of 4 rows by 8 columns requires some logic to maintain the pressed notes, that is, once the key event occurs, the note got to be played. Some bookkeeping has to be made, reflecting the fact that it has to be remembered that it had been played. As soon as it disappears from the scan balance, it got to be muted and taken out of the momentarily active notes. etc. etc.

Is this bookkeeping algorithm available somewhere? I'm aware, that reflecting velocity adds some level of complexity.

n is the "block" number, which translates to

0 = CH 10

1 = CH 1

...

9 = CH 9

A = CH 11

...

F = CH 16

It shows

00 at n!=0

01 at n=0

Does this mean that channel 10 (n=0) CAN NOT be assigned as a non-percussion part?  I am testing a file with the SYSEX command to set CH 10 to non-percussion, but my software is still playing it as percussion.

I had a catch-up with Athan Billias recently, regarding WebMIDI.
It started out as a chat about my a new App I'm developing, called MIDIWeb ( removed link ) - a browser focused on making WebMIDI practical on iOS.
It ended up with a suggestion of:

'You should start a community-driven site for cool WebMIDI sites to help spread the word and gain some momentum in preparation of WebMIDI 2.0 support... and maybe see if there's appetite for a WebMIDI working group..."

So this post is an attempt to get a little momentum going in the open:

1) Is there interest here in forming a lightweight “WebMIDI Working Group” (or special interest group) focused on:

• increasing awareness + practical adoption of WebMIDI
• sharing best practices / dev pitfalls / security + permissions gotchas
• helping keep energy behind "WebMIDI + MIDI 2.0" discussions (and feeding that back into the right standards channels)

2) Do initiatives like this already exist that I’m missing?

• I’m aware the old W3C Web MIDI Community Group is closed, and there are broader audio groups — but I’m specifically wondering if there’s an active, practical community hub today for WebMIDI work?

As a first step, I’ve put together a community-driven directory of WebMIDI sites/tools, so people can quickly see what’s possible and add their favourites:

   MIDIWeb Hub: <a href=" removed link "> removed link

This directory is now also the home page of MIDIWeb, but the directory itself is meant to be useful regardless of what browser/app you use. I’d love for it to become a “living list” that the community actually maintains.

If a WebMIDI Working Group does sound worthwhile, a few possible starter outputs could be:

• a curated "Top WebMIDI examples" shortlist (education, performance, tools, games, utilities) - perhaps piggybacking on MIDIWeb-Hub?
• a simple "Getting started" + "Known issues across platforms", "WebMIDI Learning resources"  doc
• a wish-list / priority list for “WebMIDI + MIDI 2.0” features that developers would actually use
• coordination with any existing MIDI 2.0 / education efforts (so WebMIDI doesn’t become an island)

If you’re interested, reply with:

• "Yes, I’d join" + (and what you’d want it to achieve)
• any existing groups/threads/people I should connect with?
• any WebMIDI sites/tools you think absolutely should be in the directory?

The goal here is just to get the ball rolling and see if there’s real signal before we over-organise anything! 🙂 

Cheers,
Ant, 5of12
JOIN THE MIDIWEB TESTFLIGHT BETA: removed link