asv.results

Classes

Results

Manage a set of benchmark results for a single machine and commit

Functions

iter_results_paths(results)	Iterate over all of the result file paths.
iter_results(results)	Iterate over all of the result files.
iter_results_for_machine(results, machine_name)	Iterate over all of the result files for a particular machine.
iter_results_for_machine_and_hash(results, ...)	Iterate over all of the result files with a given hash for a
iter_existing_hashes(results)	Iterate over all of the result commit hashes and dates and yields
get_existing_hashes(results)	Get a list of the commit hashes that have already been tested.
get_result_hash_from_prefix(results, machine_name, ...)	Get the 8-char result commit identifier from a potentially shorter
get_filename(machine, commit_hash, env_name)	Get the result filename for a given machine, commit_hash and
_compatible_results(result, result_params, params)	For parameterized benchmarks, obtain values from result that
format_benchmark_result(results, benchmark)	Pretty-print a benchmark result to human-readable form.
_format_benchmark_result(result, benchmark[, max_width])	Format the result from a parameterized benchmark as an ASCII table
_format_param_value(value_repr)	Format a parameter value for displaying it as test output. The

Module Contents

asv.results.iter_results_paths(results)

Iterate over all of the result file paths.

asv.results.iter_results(results)

Iterate over all of the result files.

asv.results.iter_results_for_machine(results, machine_name)

Iterate over all of the result files for a particular machine.

asv.results.iter_results_for_machine_and_hash(results, machine_name, commit)

Iterate over all of the result files with a given hash for a particular machine.

asv.results.iter_existing_hashes(results)

Iterate over all of the result commit hashes and dates and yields commit_hash.

May return duplicates. Use get_existing_hashes if that matters.

asv.results.get_existing_hashes(results)

Get a list of the commit hashes that have already been tested.

asv.results.get_result_hash_from_prefix(results, machine_name, commit_prefix)

Get the 8-char result commit identifier from a potentially shorter prefix. Only considers the set of commits that have had results computed.

Returns None if there are no matches. Raises a UserError if the prefix is non-unique.

asv.results.get_filename(machine, commit_hash, env_name)

Get the result filename for a given machine, commit_hash and environment.

If the environment name is too long, use its hash instead.

asv.results._compatible_results(result, result_params, params)

For parameterized benchmarks, obtain values from result that are compatible with parameters of benchmark

class asv.results.Results(params, requirements, commit_hash, date, python, env_name, env_vars)

Manage a set of benchmark results for a single machine and commit hash.

api_version = 2

_params

_requirements

_commit_hash

_date

_results

_samples

_stats

_benchmark_params

_profiles

_python

_env_name

_started_at

_duration

_benchmark_version

_env_vars

_stderr

_errcode

classmethod unnamed()

property commit_hash

property date

property params

property env_vars

property started_at

property duration

set_build_duration(value)

set_setup_cache_duration(setup_cache_key, value)

property benchmark_version

property stderr

property errcode

get_all_result_keys()

Return all available result keys.

get_result_keys(benchmarks)

Return result keys corresponding to benchmarks.

Parameters

benchmarksBenchmarks

Benchmarks to return results for. Used for checking benchmark versions.

Returns

keysset

Set of benchmark result keys

get_result_value(key, params)

Return the value of benchmark result.

Parameters

keystr

Benchmark name to return results for

params{list of list, None}

Set of benchmark parameters to return values for

Returns

value{float, list of float}

Benchmark result value. If the benchmark is parameterized, return a list of values.

get_result_stats(key, params)

Return the statistical information of a benchmark result.

Parameters

keystr

Benchmark name to return results for

params{list of list, None}

Set of benchmark parameters to return values for

Returns

stats{None, dict, list of dict}

Result statistics. If the benchmark is parameterized, return a list of values.

get_result_samples(key, params)

Return the raw data points of a benchmark result.

Parameters

keystr

Benchmark name to return results for

params{list of list, None}

Set of benchmark parameters to return values for

Returns

samples{None, list}

Raw result samples. If the benchmark is parameterized, return a list of values.

get_result_params(key)

Return the benchmark parameters of the given result

remove_result(key)

Remove results corresponding to a given benchmark.

remove_samples(key, selected_idx=None)

Remove measurement samples from the selected benchmark.

add_result(benchmark, result, started_at=None, duration=None, record_samples=False, append_samples=False, selected_idx=None)

Add benchmark result.

Parameters

benchmarkdict

Benchmark object

resultrunner.BenchmarkResult

Result of the benchmark.

started_atdatetime.datetime, optional

Benchmark start time.

durationfloat, optional

Benchmark total duration in seconds.

record_samplesbool, optional

Whether to save samples.

append_samplesbool, optional

Whether to combine new samples with old.

selected_idxset, optional

Which indices in a parametrized benchmark to update

_mk_pstats(bytedata)

get_profile(benchmark_name)

Get the profile data for the given benchmark name.

Parameters

benchmark_namestr

Name of benchmark

Returns

profile_datapstats.Stats

Profile data

get_profile_stats(benchmark_name)

has_profile(benchmark_name)

Does the given benchmark data have profiling information?

save(result_dir)

Save the results to disk, replacing existing results.

Parameters

result_dirstr

Path to root of results tree.

load_data(result_dir)

Load previous results for the current parameters (if any).

classmethod load(path, machine_name=None)

Load results from disk.

Parameters

pathstr

Path to results file.

machine_namestr, optional

If given, check that the results file is for the given machine.

rm(result_dir)

classmethod update(path)

property env_name

classmethod update_to_2(d)

Reformat data in api_version 1 format to version 2.

asv.results.format_benchmark_result(results, benchmark)

Pretty-print a benchmark result to human-readable form.

Parameters

resultsResults

Result set object

benchmarkdict

Benchmark dictionary

Returns

info{str, None}

One-line description of results

details{str, None}

Additional details

asv.results._format_benchmark_result(result, benchmark, max_width=None)

Format the result from a parameterized benchmark as an ASCII table

asv.results._format_param_value(value_repr)

Format a parameter value for displaying it as test output. The values are string obtained via Python repr.