Turn raw benchmark results into text and figures
benchmark_analyze(
branches = c(branch_get_or_fail("GITHUB_BASE_REF"),
branch_get_or_fail("GITHUB_HEAD_REF")),
names = NULL,
ci = 0.95
)
The names of the branches for which analysis should be created.
The names of the benchmarks to analyze. If NULL
, all
benchmarks with the branches
are taken.
The confidence level, defaults to 95%.
A character vector that summarizes the benchmarking results.
Creates two side effects:
Density plots for each element in branches
are written to
touchstone/plots
.
A text explaining the speed diff is written to
touchstone/pr-comment/info.txt
for every registered benchmarking
expression. See vignette("inference", package = "touchstone")
for
details.
Requires dplyr::dplyr, ggplot2::ggplot2 and glue::glue.