The short version: xterm -xrm 'xterm*eightBitInput: false' yields xterm: fatal pty error 23 (errno=22) on tty /dev/pts/4
on at least the stock Debian PPC kernel (2.4.4), the latest BitKeeper tree (2.4.10-pre8), and 2.4.9-benh0. This problem does not occur on the i386 architecture. The somewhat longer version: I have the 'eightBitInput: false' resource set in my .Xresources, so that when I ssh to other machines, meta works properly in xterm. (Yes, yes, I realize that 'metaSendsEscape: true' will do this too, but I want Alt, Super, Hyper, etc to work as well.) This does not work with any linux-ppc kernel that I have tried; I get the error above. errno 22 is EINVAL, and 23 refers to the error occuring on a TIOCSETP call (which looking at the xterm code could also mean that a TCSETA call failed). I've tried to debug this, but I started looking at the xterm code and kernel termios and my brain leapt out through my ear and is currently cowering under the bed. I've been trying to lure it out with treats, but have thus far been unsuccessful. The only notable change that this Xresource should cause is setting CS7 instead of CS8 in the termios control mode flags, in a TCSETA call. I have been unsuccessful in trying to replicate this failure in a small test program, however. Has anyone else encountered this? For that matter, why is termbits.h different from i386 to ppc in the same kernel? I can't think of any plausible reason. --Jered [EMAIL PROTECTED]
