Benchmark chunks — benchmark • chunkhooks

hook_benchmark sets a hook to benchmark chunks with the benchmark=TRUE option. The name of the trigger chunk option can be changed via the chunk_option parameter. The result is printed right after chunk outputs. See examples for the default printing format by format_benchmark.

hook_benchmark(
  trigger = "benchmark",
  default = NULL,
  format = format_benchmark,
  .set = TRUE
)

format_benchmark(result, options)

benchmarks

Arguments

trigger	A name of chunk option that triggers benchmark (default: `"benchmark"`). In order to trigger benchmark, specify `TRUE` as a value of this option.
default	A default value for the chunk option that `trigger`s the hook (default: `NULL`).
format	A function to format a benchmark result (default: `format_benchmark`). It must accept two arguments, where the first is the benchmark result and the second is the list of current chunk options. `NULL` suppresses printing.
.set	`TRUE` or `FALSE` to set the hook.
result	A result of benchmark
options	A list of current chunk options

Format

An object of class environment of length 0.

Value

invisible hook function

Details

benchmarks records the results of benchmarks from chunks as a list named by chunk labels. If one requires complex formatting of benchmark results, then suppress automatic formatting by hook_benchmark(format = NULL). Then, retrieve benchmark results from benchmarks[["chunk-label"]]. Furthermore, format can happen conditionally by utilizing current chunk options via the second argument of the formatting function.

Examples

# Set a hook that triggers benchmarks if the `time` chunk option is not `NULL`.
hook_benchmark("time")

# Example of the default output format
# Input is sec. Output is prettified.
format_benchmark(1234, options = list(label = "example-chunk"))
#> [1] "example-chunk: 20m 34s"