Skip to content

MON 1B on device 0 - the command line is CR-terminated (read side + emulator)

Companion to README.md, which carves the MON 1B dispatch and worker. That README's honest caveat says "the wait/no-wait and logical-device-0 behaviours come from the manual." This note closes the logical-device-0 half of that gap with byte + runtime evidence, from two independent directions.

One-line answer: a program reading logical device 0 (the SINTRAN command buffer) via MON 1B gets the invocation command line terminated by CR (015B / 0x0D) - never 47B, never raw End-Of-File. A program launched with no arguments still reads a lone CR.

All addresses octal unless prefixed 0x. Full paths are absolute so this is readable cold.

Two sentinels, two layers (do not conflate them)

Value Layer Who sees it
47B (0x27, ') command-processor source string terminator the command PROCESSOR, on the write/prepare side. A device-0 reader never sees it.
015B (0x0D, CR) the device-0 line terminator the READER, via MON 1B
-1 / 177777B the level-4 device "no byte available" sentinel the INBT worker internally (see README)

These are three different things. Conflating the first two nearly caused a wrong experiment (parking the command byte-pointer on 47B and breaking in the level-4 read loop watches the wrong sentinel).

WRITE side - VERIFIED (command-processor bytes, L07)

The command processor reads the source line and, at the 47B source marker, substitutes CR and resets the command byte pointer (CPNT):

050773: SAA 15        % A := 015B (CR)
        ... SBYT      % store that byte into the command buffer
        ... reset CPNT

So 47B is an internal SOURCE marker; what lands in the command buffer for a MON 1B reader ends in CR. (Trace captured while watching CPNT = 144033B; the command buffer machinery is CPNT byte pointer + CSTRIN = 144035B string - NOT CBUF = 170207B, which is a boot-time scratch pointer, INTEGER CBUF % ADDRESS OF CURRENT I/O BUFFER, /mnt/e/Dev/Ronny/NDInsight/SINTRAN/NPL-SOURCE/NPL/PH-P2-START-BASE.NPL:24.)

READ side - VERIFIED (runtime, on the ND-500 emulator)

The ND Linker linker-b01.dom is a real program that reads device 0 past its line, so it answers the read side directly - no ND-100 guest required. Under the nd500x emulator, with device 0 delivering CR:

  • its first device-0 read (MON 1B INBT @ PC = 0xB004E759) returns 0x0D;
  • it then prints its banner + startup dialogue and never reads device 0 again.

So a device-0 read yields args + CR, and a reader consumes that and stops.

Emulator consequence - VERIFIED (measured)

Returning raw EOF at end-of-line instead of the CR makes that same linker busy-spin: 20,088 consecutive MON 1B INBT at PC = 0xB004E759 within 600k instructions, no other activity. Delivering the CR lets it complete its line and proceed. Fixed in nd500x at /home/ronny/repos/nd500x/src/libmon/mon_file_table.c (mon_read_command_buffer_char): return the stored bytes, then exactly one CR if the line did not already end in one, then end-of-line.

Still UNPROVEN (do not guess)

What a device-0 read returns past the CR (a read after end-of-line). The linker never does it, so there is no runtime evidence, and the deciding code is the level-4 byte-input worker whose device-0 branch is not isolated (see README - M1 = 071633B is a two-word stub M1: "INBT"; GO IOB14; the behaviour lives in the level-4 worker INBT, live- proven as resident code at 032471B but its device-0 path not yet traced). The nd500x handler currently treats a past-CR read as a process suspend, which is defensible but not proven.

Cross-references

  • Dispatch + worker carve: README.md
  • Manual contract (device 0 = command buffer, "parameters following the program name", break and echo both 1): /mnt/e/Dev/Ronny/NDInsight/Developer/MON/calls/1B_InByte.yaml
  • MON 12B SETCM ("the command buffer contains the last command input from the terminal"): /mnt/e/Dev/Ronny/NDInsight/Developer/MON/calls/12B_SetCommandBuffer.yaml
  • NPL answers (device-0 / command buffer / RSIO / DESCF): /mnt/e/Dev/Ronny/NDInsight/SINTRAN/ND500/NPL-ANSWERS-DEVICE0-CMDBUF-RSIO-DESCF.md
  • Emulator fix + read-side trace: /home/ronny/repos/nd500x commit 19f7354