I have noticed that if you run a C program for the ti 84 plus ce, and press the on key while the program is running, and then quit the program normally, the program quits normally, but there is the os message that shows an error: break message, just like you just force-stopped a ti-basic program with the On Key. I tried disabling on-latching and clearing the on-latch here but that didn't seem to work: https://github.com/randomguy70/SmartNotesCE/blob/1490892276cb0c432a0fa111fdfe560c1bf39ebb/smart_notes_ce/src/main.c#L12.
I then found that professionally done programs such as Hexaedit had the same issue, so I assume that this is a common problem. Is this a bug, or is it fixable with a simple function, and if so, what function? If you have any insights, please comment below. Thank you in advance!

I think it's something hard-coded into the OS. I'm guessing that hypothetically, if we flashed a custom OS to the calculator, we could disable that behaviour.

Use this:

Code:


For anyone wondering how I even figured this out, the OS uses the main set of interrupts, while keypadc uses the FIQ set. So, replacing the address of the FIQ interrupts in the define for kb_ClearOnLatch with the address of the main set resets the OS's copy instead.

EDIT:
Added that to the cleanup code for the toolchain, so in future releases programs will just do that by default.

Shouldn't keypadc do that cleanup instead of the CRT? Especially since there's already kb_Reset(). Seems like the cleanup should go in kb_Reset() and change the documentation to say that it's required to prevent the ERR: BREAK issue.

It's not an issue with keypadc specifically, it's just that GetCSC resets the interrupt. So, you could have the same issue in a program which calls GetCSC, then waits for a while without calling it before exiting. So, I figured that the crt was a better location for it, since it's not specifically related to keypadc. Also, requiring users to call a function to fix a hard-to-catch issue would likely just result in lots of people not fixing the issue.

Thank you, it works now!