Once you've run evo init, there's no supported way to update the stored benchmark command, metric direction, or target. If the --benchmark string was typoed (say, missing a {worktree} placeholder), the options right now are:
evo reset --yes and re-init, which blows away any experiments already created- Hand-edit
.evo/run_NNNN/config.json directly
Both are bad.
Would be nice to have:
evo config set benchmark "..."evo config set metric minevo config set target <path>evo config show
Hit this a couple times — usually it's a mistake in the initial --benchmark string that only surfaces when you try evo run and the path doesn't resolve. By that point there's already a baseline experiment you don't want to throw out.
Once you've run
evo init, there's no supported way to update the stored benchmark command, metric direction, or target. If the--benchmarkstring was typoed (say, missing a{worktree}placeholder), the options right now are:evo reset --yesand re-init, which blows away any experiments already created.evo/run_NNNN/config.jsondirectlyBoth are bad.
Would be nice to have:
evo config set benchmark "..."evo config set metric minevo config set target <path>evo config showHit this a couple times — usually it's a mistake in the initial
--benchmarkstring that only surfaces when you tryevo runand the path doesn't resolve. By that point there's already a baseline experiment you don't want to throw out.