Project: The Sound of Violence.
I’m still bedevilled by the serial “xmit-stopped when CTS floating” issue. I should ignore it for now and construct the software simplex state-machine for the PedalBox, but I’m hooked on the problem now…
Inspecting what tip and seyon do was not very illuminating (my software using same port modes, termios flags and file-decriptor flags but still suffering different write behaviour).
One particular factor that intrigued me was that both tip and seyon use separate processes for reading and writing the serial-port, and in the case of tip, a pipeline between them to gather the bytes. Seems way more complicated than necessary, unless it *is* necessary to avoid the CTS problem. Did I miss something?
So, now the plan is to download and install a much simpler direct serial-port terminal-emulator program – minicom – and see what *that* does (using truss to trace the system-calls it makes). I strongly suspect that it does not fork multiple processes, but either way it should provide good clues.
A long time ago, on much simpler hardware, a 386SX ISA PC, minicom was my preference for serial comms, but that was in the days of dial-up BBS!
The other possibility is that CTS-sensitivity is a hardware limitation of the Zilog 8530 asynchronous serial controller (UART chip). That suggests that trying my program under Solaris on an Ultra-10 (Siemens 82532 UART) and a Toshiba Satellite 300CTD (NatSemi 16550A UART). We’ll see…
PS: Retro Cammage
Now all I need is a video camera!