v0.16.0

Breaking Changes

• Change mch:noop to _
• Remove function_benchmarks option

Improvements

• Add raw scores to results
• Sort results by name

v0.15.0

New Features

• Add a new command _ <string>, which takes a string and returns 0
  • This is useful for benchmarking macros

v0.14.1

Fixes

• auto_start is not working

Improvements

• Print the benchmark name before manual start

v0.14.0

New Features

• Benchmarks can now be started manually via the auto_start option #56
  • You can start the benchmark by executing /function #load after the server has been initialized
• Output file name can now be customized via the output option #60

Changes

• #mch:setup and #mch:teardown are now executed once per group
• @benchmark annotation can now be written on lines other than the first

Fixes

• #mch:setup.trial and #mch:teardown.trial are executed on the dry run #73
• #mch:setup is executed only once #74

v0.13.1

Bump version to 0.13.1

v0.13.0

Breaking Changes

• The execute_benchmarks option now takes commands instead of function names

Improvements

• Add samples
• The baseline (empty function) is now always benchmarked #57

New Features

• You can now specify the formats of the results with the formats option (json or md) #27
• The #mch:teardown fixture #67
  • Executed at the end of the last fork of the entire benchmark
• The new execute_benchmarks option for benchmarking a signle command #68
• The function_benchmarks option that allows more stable function benchmarks #65
  • By setting the pack.mch of the data pack's pack.mcmeta to true, the data pack is recognized as for benchmarking
  • By adding # @benchmark to the first line of a function, the function is recongnized as for benchmarking #69
  • Among the benchmarking data packs, only one being benchmarked is enabled, and the functions it contains are benchmarked

Fixes

• #mch:setup was executed inappropriately #70

v0.12.0

Breaking Changes

• Changed the case of command-line options to snake_case #66

Improvements

• Scores reaching maxCommandChainLength are no longer included in the results #36
• Now displays total progress #48

v0.11.0

Improvements

• Generate data pack only when necessary #61
• Validate options #59

New Features

• mch can now be used as a library (with only minimal features at the moment) #52

Internals

• Improved stability during development with the test workflow

v0.10.0

Breaking Changes

• parsing-benchmarks now takes commands instead of function paths

New Features

• Include Minecraft version in mch-results.json #58

Internal Changes

• Use spy for code injection

v0.9.1

Fixes

• Fixed an error when mch-config.json is missing #54
• Fixed an issue with the customized server path not working #55