Config Tool for Mac user (hopefully a temporary post!!)

[elrules]

I have to investigate (as Sysex Librarian is open sourced) how Sysex Librarian works. Current version has an interesting feature to act as a MIDI device, to connect to other midi ports. But it is only for input (or only for output, I don´t remember now).

The possible solution using Sysex Librarian would be to build a new Sysex Librarian, able to both send and receive, and find a way to send the received data not with MIDI, but with other kind of communication to a Java program.

By the way, I have managed to install Mac OS X to my Intel-based laptop. No virtual machines. So I can fully test MCT in Mac. I have also installed XCode to be able to program in objective-c. Let's see what I can do with that...

[ole_hansen]

keep your spirits high and keep up the good work elrules!

[elrules]

Quite disappointed with the performance of the Java MIDI implementation in MacOS.

I have being finding out were MCT is fucked and I sort of found something (weird):

Let's talk only about Miscellaneous config. To focus on something.

-If my first action after I open the midi ports is that I load misc settings a request is sent. Both the request and the reply are received from megadrum. Settings are loaded successfully. I can load all the times I want and it works ok.
If then I click on Send misc settings to MD, it sends the corresponding sysex, and at this moment the MIDI system is fucked. I cannot load or send anything (load or send commands). Weird...

-If my first action after I open the midi ports is that I send Misc settings to MD, a sysex is sent, MD settings are updated, and the same sysex is received through the input. That´s normal. I can repeat this action without problems
If I now load Misc settings again everything is fucked. I have to close the application.

Not sure if every time, but, when the MIDI system is fucked and cannot load or send anything, if I close the application and start it again, I can then send again. That probably means that MD is not affected by this weird behavior, so it should be all fault of the poor MIDI system in Mac.

I have tested both Mandalone and MMJ libraries. Both of them behave the same. So it seems like something inside the java platform.

I have to find another fresh way to test things or another way to solve this crap.
Ideas are well received

[ruffneck]

Hi,
i experience exactly the same as described above. So i tried the following:

Installed Sun VirtualBox (Free virtualpc software) and configured and installed Ubuntu Linux on it. Now you can share your physical USB ports with VirtualBox so i thought this would be perfect!

So once i got Ubuntu up and running i patched it, installed Sun Java 5 and the VirtualBox drivers. Rebooted.
Installed MCT and......

....exactly the same results as running with OS X.

At this point i'm starting to think that either the USB part on OS X is fucked or the javamidi on OSX is fucked or the linux javamidi is fucked.

I'll see if i can install Windows XP on the VirtualBox and see if that works. I'll report back as soon as i have something to report.

And guess what. It works within a VirtualBox virtualmachine loaded with Windows XP! The host is still OS X.

/Kari

ps. sorry for the bad language, but this is getting frustrating.

[davem]

Would it be worth me writing a small native application to try this using the CoreMidi framework directly in C independently of MCT and check that it's not horribly broken in some way? Hopefully my Synthex kit should be here soon so I can help to test/diagnose the problem.

Edit: Scratch that, just saw the bit about Sysex Librarian. I'll try and chip in some when my PCB arrives.

[davem]

So, my v3.1 board arrived on Monday, it's all soldered up and working great, so I've had some time to fiddle with MCT on my aluminium Macbook. I've installed MMJ and am using the 32-bit 1.5 JRE under OS X 10.5.7.

What can I say... loading and saving misc settings with the stock 1.71 version of MCT is working relatively fine! I've tried all the scenarios that elrules has set out. For example, I tried the following:

• Load misc settings from MD
• Set big VU meter in MCT
• Save misc settings to MD

and the big VU meter appears on the LCD. Unticking the box in MCT and reloading settings from the MD causes the box to be reticked. I'm a bit limited in what else I can test for the time being since my pad is difficult to hook up, but this afternoon I should have some rotary encoders turn up so I will construct a keyboard to test that, for example, pad settings are being set correctly.

However I can still make the app fall over. If I continuously write pad information to the MD, then after about 10-15 writes the application log blanks out and the status bar hangs at "writing pad settings to md". Edit: After opening MCT up in NetBeans and debugging, it looks like this issue is to do with the application log, or at least the log text. Commenting out the updateGUI() method in ApplicationLog prevents the hang and loading/saving MD values works fine.

I'm unable to describe the discrepancy others have been seeing with MCT + OS X. The only obvious difference on the boards besides number of inputs is that the previous designs use PIC18F2550 for USB whereas my 3.1 has a PIC18F14K50. But to be honest, I'm no electrical engineer so I couldn't say whether this has any impact on the final outcome or not.

Please let me know if there's anything else I can test out.

PS: I also tried 1.71 out on my Boot Camp partition which has Windows XP installed on it and it also works great there.

[elrules]

Can you load/send all settings? (it doesn't matter if you don't have pads as configs are shown in the LCD)

[davem]

Yup, works fine. In the middle of building my encoders at the moment so hopefully I will be able to do a bit more testing soon.

[ole_hansen]

so does this mean that the new 3.1 board works?
a new update works?
or that OS X 10.5.7 works?

[fuzzysnuggleduck]

Just some updates here for the Mac people on my latest FW upgrade and MCT testing on Mac.

I've got a v2.7 Synthex board and the following applies to both my friends 10.5.3 install and my 10.6.1 install:

1. Updating FW via MCT 1.83 doesn't work.
2. Updating FW via SysEx Librarian doesn't work.

Now, maybe I'm just stupid but I couldn't get it to work after about 2 hours of trying both MCT and SysEx Librarian. I was able to update the firmware today at work with a Windows Server 2003 box and MCT 1.83 so at least I have the newest firmware with all the goodies I've been missing for some time now. Maybe I'm really just stupid and didn't get the steps right yesterday but it worked in fairly short order today with a Windows machine.

MCT 1.83 doesn't work for sending or receiving any settings from the MegaDrum with the latest firmware (20091101) and Snow Leopard. MMJ is installed. Java is set to 32-bit (only one available in "Java Preferences"). I have yet to try MCT 1.83 send/receive to the new firmware with Leopard (friends 10.5.3 install).

I can however seem to capture the SysEx messages when I use "Send ConfSysex" on the MegaDrum and SysEx Library to capture the data. I don't know what the data is supposed to look like, but SysEx Library receives 105 messages, so I think that's a good sign that SysEx messages are working to some degree or another.

Edit: I just tried using an older version of 32-bit Java on Snow Leopard (1.5) and still no success with MCT 1.83 sending or receiving any settings.

Edit 2: I tried creating sysx files from MCT and then loading them onto the MD with SysEx Librarian on Snow Leopard... that also doesn't work.