DrWatson

This small configuration cache exercises production followed by reuse:

From examples/kitchen-sink in the source checkout, run julia setup.jl extras. Then start Julia with julia --project=.controller/extras for the code below.

using PerfChecker, DrWatson
mktempdir() do directory
    calls = Ref(0)
    parameters = Dict("recipe"=>"cache_roundtrip", "revision"=>1)
    producer = p -> begin
        calls[] += 1
        Dict("parameters"=>p, "measurement_executed"=>false)
    end
    first_result, first_path = drwatson_produce_or_load(producer, parameters;
        directory, tag=false)
    second_result, second_path = drwatson_produce_or_load(producer, parameters;
        directory, tag=false)
    @assert first_result == second_result && first_path == second_path
    println((calls[], second_result["measurement_executed"]))
end

Observed output: (1, false). The producer ran once; the second call reused the saved 6,784-byte cache. The receipt retains the cache hash. This proves configuration reuse, without claiming a new measurement or publishing a configuration cache as performance evidence.

Cache one measured package feature

From examples/kitchen-sink:

julia setup.jl extras
julia --project=.controller/extras drwatson.jl run datastructures

The supplied script uses drwatson_parameters and drwatson_savename on the suite plan, then drwatson_run_suite for one latest-release heap feature. It asserts result["passed"] before printing its cache filename. Set PERFCHECKER_FORCE=true only when intentionally remeasuring. There is no separately recorded measured package-cache run in this example.