Headless runs

Most of what we build is infrastructure, with no front end to open. The demonstration is the execution itself, recorded in a terminal: a device and the host writing the same bytes, a power cut at every byte, and one C surface on two architectures.

• The device and the host agreeA RISC-V device tier and the host write the same volume, byte for byte: 0x299bd5d0e1821fab on both.Open the run →
• A power cut at every byteThe log is cut at each byte of its last record and every sixty-fourth byte after it. Every run opens, and none finds a record that was not there.Open the run →
• A record round tripThirteen steps, eight records, one digest: the same run the device tier converges on.Open the run →
• One test body, two architecturesThe same C test body linked from C and run under QEMU on a Cortex-M4 and a RISC-V core, with the same fifteen measurements on both.Open the run →
• Three refusalsA medium that refuses a write, a worn medium, and a channel that holds a write past the end of the run.Open the run →