Description

The tutorial attempts to use logmetric to log entire arrays, but the function only accepts single Float64 values.

Location

• docs/src/tutorial.md:98

Issue

Tutorial shows:

logmetric(mlf, exprun, "pricepath$(idx)", pricepath)

Where pricepath is an Array{Float64}. However, all logmetric method signatures only accept single Float64 values:

logmetric(instance::MLFlow, run::Run, key::String, value::Float64; timestamp, step)

Impact

• Users following the tutorial will encounter MethodError: no method matching logmetric(..., ::Vector{Float64})
• The tutorial fails when trying to log the main experimental results
• This breaks the core functionality demonstration

Expected Behavior

Either:

1. Tutorial should log individual values from the array in a loop, or
2. Tutorial should use a different approach (like logbatch with multiple metrics), or
3. Implement array support in logmetric (though this may conflict with MLFlow's metric concept)

Suggested Fix

Replace the problematic line with:

# Log each point in the price path as a separate metric with step parameter
for (step, value) in enumerate(pricepath)
    logmetric(mlf, exprun, "pricepath$(idx)", Float64(value); step=step-1)
end

Actual Behavior

Tutorial shows code that causes type errors due to array vs scalar mismatch.

Environment

• MLFlowClient.jl version: 0.6.0
• Julia version: 1.11.6

Found via systematic API analysis using methods(logmetric) to verify method signatures against tutorial usage.