A response-time bound is only as good as the execution times fed to it. If those came from an oscilloscope, the bound means something. If they came from a spreadsheet, the bound is arithmetic about a guess — and the output looks identical either way.
Nothing in the current toolchain forces that distinction to be visible. An analysis should carry, alongside every input, whether it was measured and under what conditions.
Nothing is implemented. The measurement side needs hardware this project does not have, and the honest sequence is to define the format first and populate it when a fixture exists.
Publishing a measurement format before having measurements is the right order. Publishing measurements before having hardware is not.
dy-wcet is the piece that works
today: response-time analysis for fixed-priority task sets, in integer
arithmetic that refuses rather than rounds.
Apache-2.0 OR MIT, at your option.