R code

R runs through WebR, R compiled to WebAssembly. It is the language to use when the check you care about is statistical rather than structural: response-time distribution, error rate, rolling averages.

The API

Symbol	Role
input	The value received on the input port
vars	The run variables
cat(...), print(...)	Write to the execution log
The last expression	Its value leaves on the result port

Scalars are unwrapped; a named list comes back out as a JSON object.

Example

temps <- unlist(input$temps_ms)

moyenne <- mean(temps)
p95 <- quantile(temps, 0.95)

cat(sprintf("moyenne %.1f ms, p95 %.1f ms\n", moyenne, p95))

list(moyenne = moyenne, p95 = unname(p95), n = length(temps))

Wire the output onto an lt Assert to fail the run if the p95 exceeds your budget.

External packages

The packages field accepts CRAN packages, served from WebR’s WebAssembly repository.

Things worth knowing

The R engine is a shared singleton, whose runs are queued. A long R script therefore delays the others in the same run — keep them short, or raise the timeout accordingly.