Could someone rerun your DOS demo? A six-byte test
· Compsyscon preservation editorial desk · Worked example and worksheet
Two runs produced a file containing BLUE. A third produced GOLD. All three files were six bytes long. A note saying “CARD.TXT appeared and its size was correct” would mark every run as a match. The file contents tell a different story.
1. Record the setup someone else will need
We used DOSBox Staging 0.83.1, build 8888e, on an arm64 Mac running macOS 27.0.1. Each run started in a fresh directory containing two input files. We disabled the primary and local configuration files and automatic startup commands with --noprimaryconf --nolocalconf --noautoexec. We set cpu_cycles=3000 and window_size=640x480. These settings describe the test we ran.
2. Run the same two files in two fresh directories
Save these two lines as DEMO.BAT in plain ASCII. End each line with a DOS carriage return and line feed, including the last one.
@echo off
type LABEL.TXT > CARD.TXTFor LABEL.TXT, save the four letters BLUE followed by the same two line-ending bytes. Mount the folder as drive C in the emulator, switch to C and run CALL DEMO.BAT. The batch file writes its result into the mounted folder. Start the second run with fresh copies of just the two inputs. A file left from a previous run could otherwise look like a new result.
3. Keep the procedure and change one input
In the third directory, replace BLUE with GOLD in the label and keep the line ending. The first two outputs matched byte for byte. The third had the same length, but its SHA-256 began 14a1e4ae instead of c319eb41. The three-run record includes the full hashes for independent comparison. If BLUE produces a different hash on your system, check the input bytes and line endings before changing emulator settings.
4. Say exactly what the comparison establishes
This demo exercises a file read and write. It supplies no evidence about the speed, sound or graphics of an original machine. An exhibit that depends on any of those needs its own observable check. Keep the donated machine’s provenance record beside the demo instructions so a future curator can tell which observations came from the physical machine.
To check whether a copied output still has the expected bytes, follow the worked checksum comparison.


