That's embarrassing. JUCE did change the internal handling for comboboxes in the last version. Should be fixed now.

I'll need to read the MPE specs to get a full image of how it works, but internally I am using an own message format which allows a little bit more information than just the pure MIDI messages (eg. 256 MIDI channels etc). I think adding MPE support shouldn't be too hard. It's a kind of niche thing right now (apart from Bigwig I don't know any host that supports this yet), but in the future it will become interesting.

But if I understand you correctly, even with the Linnstrument it's only one active note per MIDI channel? If that's the case, my assumption is still correct, I'll just need to add the storage for different channels (which I need to do anyway for any sort of multitimbral sampler).

Yeah, let me know how it works on Linux.

I uploaded the new build version 647 anyway, so you don't have to build it for Windows.

Aha! I just moved my external files and it loaded up no problem. It was crashing as soon as I opened the preset, no clicking on anything. So my next task is to narrow down which file is causing the crash and then which bit of code.

Yes I introduced this glitch when I rewrote the voice count system but I think I already fixed it again (but I may have not uploaded the most recent changes).

This issue seems to be fixed now :)

The Synth.getAttribute() method is working, but you need a getAttribute() method for connected modules, which was indeed missing. I added it for all module types.

BTW, it is a bit more efficient (and more compact) to write toggle behavior like this:

MidiMuter.setAttribute(0, 1.0 - MidiMuter.getAttribute(0));

because you'll get away without branching...

Yes this is a weird one. I also encountered it a while ago but I was too lazy to fix it - and I thought it was my crappy computer acting up :)

Thanks for spotting this. I'll fix it for the next build. In the meantime you can just do this:

TableEnvelope.setAttribute(0, Math.max(1, value));

It's online.

Thanks, but I don't think it's necessary (porting it to the forum is not that much trouble as both systems are working with Markdown)...

Just need to find the time :)

Yes the problem is that my current web hosting service doesn't support running a Node.js based forum system so I had to outsource this to another cloud server service and I am too stupid to figure out the DNS settings correctly.

In the long run I'll migrate it back to my original hosting company and then I'll have the correct domain again. In the mean time you can just type forum.hise.audio and it should redirect you here...

Thank you!

http://www.cockos.com/licecap/

more ideas:
3. fill in gaps by rounding pixel position - it would just look cleaner without having to do antialiasing. e.g. for the sliders, you just ensure there is always 1 pixel space between them. or 0 pixel space in the case of the mapper.

idea 1 - I agree this is not really needed.

@aye58sk4:

yeah, removing the sine calculation code fixes the crash.

This was a tough one. Actually the code that was responsible for the crash was this line:

ccValue = Message.getControllerValue();

The debugger gave no clues (like always if you debug multithreading issues) and the sine calculation stuff just delayed the call to this API method. If multiple CCs come in in a short time the internal message was overriden and the API call was using a deprecated pointer to the older message.

Moving this line before the sine loop also fixed the problem. But I changed the handling of the internal message so the original script (and every other script that does heavy calculating in the deffered callbacks) won't crash anymore…