This implementation of --conclude is basically identical
to the existing implementation for --prepare. I have not
adapted any existing tests. All tests which test --conclude
functionality are new.
The --conclude commands will run in the same order in which
they are supplied on the cli in the same manner as --prepare
and there must either be one --conclude command, or else the
same number of --concludes as there are commands.
This adds a new `--sort` option to choose the way in which the results
in the speed comparison and the markup exports are ordered.
closes#614closes#601
This adds a new "raw executor" (next to the "shell executor") that
allows hyperfine to execute commands directly without any intermediate
shell.
The command line is split into tokens (using the `shell-words` crate),
and according to POSIX rules. The first token is taken as the executable,
and the rest as arguments.
The new executor is enabled by default. In order to select the shell
executor, users will have to pass `--shell=default`.
This allows us to reduce measurement noise and to benchmark very quick
commands. It also decreases the time to run benchmarks, as we don't need
the calibration phase.
Also, it allows one to make sure that the executed command is not
implemented as a shell builtin. For example `hyperfine true`
and `hyperfine --shell=default true` return different times due
to the fact that `bash` executes `true` as a NOP.
Co-authored: Ciprian Dorin Craciun <ciprian@volution.ro>
Add a --setup option for the use-case of wanting to do one-off setup
before a set of benchmarks, not the once-per-test setup that --prepare
does. This is useful for the cases noted in the updated documentation.
Per the feedback on https://github.com/sharkdp/hyperfine/pull/448 this
new "--setup" option will steal the "-s" short option from
"--style" (initially this was called "--build" and used the
non-conflicting "-b").
Potential future work:
I'd prefer if this and --cleanup took N number of commands, so you
could do e.g.:
hyperfine -L rev master,next \
-b 'git worktree add /tmp/test-{r}' \
-b 'make -C /tmp/test-{r} all' \
'make -C /tmp/test-{r} test' \
-c 'git worktree remove /tmp/test-{r}'
I.e. a shortcut around not providing these with &&, which makes things
a bit more readable.
But the --cleanup option doesn't do that, so let's just go with what
it's doing for consistency, so for this you'll now need to do:
hyperfine -L rev master,next \
-b 'git worktree add /tmp/test-{r} &&
make -C /tmp/test-{r} all' \
'make -C /tmp/test-{r} test' \
-c 'git worktree remove /tmp/test-{r}'
Extend the --runs=N option added in d78c33b (Added options to specify
the max/exact numbers of runs., 2018-09-09) to support --runs=1
instead of dying with a usage error.
This is useful to do ad-hoc testing of your commands while they might
still have syntax errors, or just for doing one run where you don't
care about the stddev.
Before:
$ /usr/bin/hyperfine -s basic -r 1 -L n 5,10 'sleep 0.{n}'
Error: Number of runs below two
After (-s basic) also works:
$ hyperfine -r 1 -L n 5,10 'sleep 0.{n}'
Benchmark 1: sleep 0.5
Time (abs ≡): 500.6 ms [User: 0.6 ms, System: 0.0 ms]
Benchmark 2: sleep 0.10
Time (abs ≡): 100.8 ms [User: 0.7 ms, System: 0.0 ms]
Summary
'sleep 0.10' ran
4.97 ± 0.00 times faster than 'sleep 0.5'
This likewise combines correctly with -m and -M, probably not very
useful, but if you're tweaking an existing command-line:
$ hyperfine -m 1 -M 1 -L n 5,10 'sleep 0.{n}'
Benchmark 1: sleep 0.5
Time (abs ≡): 500.6 ms [User: 0.5 ms, System: 0.0 ms]
Benchmark 2: sleep 0.10
Time (abs ≡): 100.6 ms [User: 0.6 ms, System: 0.0 ms]
Summary
'sleep 0.10' ran
4.98 ± 0.00 times faster than 'sleep 0.5'
The "± 0.00" output in "faster than" should probably be adjusted too,
or we could keep it for consistency. I didn't implement that because
this is the first time I do anything in Rust, and I ran out of
template to copy when wanting to quickly implement this in
write_benchmark_comparison() in main.rs.
