From 9ef58a2e9feff18507e4b3a2b51530e2626d85db Mon Sep 17 00:00:00 2001 From: sdarwin Date: Fri, 19 Jul 2024 06:09:53 -0600 Subject: [PATCH] Docs: list bench command line options --- bench/README.md | 22 ++++++++++++++++++++++ 1 file changed, 22 insertions(+) diff --git a/bench/README.md b/bench/README.md index 68e7c8573..4e2d89864 100644 --- a/bench/README.md +++ b/bench/README.md @@ -5,6 +5,28 @@ fetch the third party repositories. Then run the bench program with no arguments for a list of command line options. +``` +Usage: bench [options...] ... + +Options: -t:[p][s] Test parsing, serialization or both + (default both) + -i:[b][d][r][c][n] Test the specified implementations + (b: Boost.JSON, pool storage) + (d: Boost.JSON, default storage) + (u: Boost.JSON, null parser) + (s: Boost.JSON, convenient functions) + (o: Boost.JSON, stream operators) + (default all) + -n: Number of trials (default 6) + -b: Branch label for boost implementations + -m:(i|p|n) Number parsing mode + (i: imprecise) + (p: precise) + (n: none) + (default imprecise) + -f Include file IO into consideration when testing parsers +``` + When building with b2, it is possible to create several different copies of the bench program for different build properties (toolset, build variant, etc.). Rather than figuring out where those programs are located from b2 output you