🚀 go-pugleaf

I have uncovered a problem with emacs on ESIX (Intel 386 SYSV). The code
installs (i.e. compiles and runs mostly right) after some #defines
have been done to allow for the fact that what emacs calls tchars,
ESIX calls tc which is declared in <sys/ttold.h>, and this brings forth
more needs for #includes of <sys/ptem.h> and perhaps a couple of other
things.

The editor works just fine except for one thing which is important for
me because I want to use it with the Boyer-Moore Theorem Prover.

When M-x shell invokes an inferior shell, commands from the keyboard
reach the inferior process OK, but stdout from the process appears to
be buffered instead of being sent to emacs at least a line at a time.

The evidence for this is that when I manage to stop the inferior process
and the shell, all the missing output appears on the emacs buffer display
at once.

I have been through sysdep, and have not found any obvious clues. ESIX
supports setvbuf(), and I have tried setting the mode in setvbuf
both for line at a time and character at a time with no change in the
results. Editing out the setvbuf() so that setbuf is used also does
not help.

I have not at this time run "cc -E" against sysdep to see if  setvbuf
calls are actually reacing the compiler.

What else -- ESIX supports PTY's and X windows, but defining HAS_PTYS
and HAS_SYSV_PTYS or things like that but actually in sysdep does no
good.

Obviously I can't ask you chaps to find the bugs. There probably isn't
and EVERX UNIX within miles. Setting the XENIX #define gave me a good
deal of grief earlier although ESIX is claimed to be XENIX compatible??.

Can you suggest where I should look in the code next. I have sdb, but
no dbx. I also have gcc and gdb, but because of the COFF_ENCAPSULATE
problem I have not tried to use them.

	Yours Hopefully (and Sincerely)

		John Gabriel (gabriel@ees.anl.gov)