1.1.0

Fixed

• Fixed wrong exit code being returned when a panic occurs outside of the function being benchmarked.
• MacOS/Windows: Fix connection issue that manifested itself in a few different ways.
• Use new version of plotters. No new features but it fixes a bug that caused criterion to
  hang indefinitely.

Added

• Load configuration options 'criterion.toml' if 'Criterion.toml' isn't available.

1.0.1

Fixed

• Changed opacity of the violin plots to full.
• Fixed violin chart X axis not starting at zero in the plotters backend.
• Fixed panic in the history report code.

1.0.0

Fixed

• Fixed potential panic if a benchmark took zero time.
• cargo-criterion now calls cargo metadata to find the path to the target directory. This fixes
  the location of the target directory in workspaces.

Added

• Added a report showing the historical performance of a benchmark.

1.0.0-alpha3

Added

• The criterion.toml file can now be used to configure the colors used for the generated plots.

1.0.0-alpha2

Added

• Initial version of cargo-criterion

Fixed

• Fixed problem where benchmarks that relied on dynamically linked libraries would fail
  in cargo-criterion but not in cargo bench.
• Sort the benchmark targets before running them. This should ensure a stable execution order
  for all benchmarks.

Added

• Added --message-format=json option, which prints JSON messages about the benchmarks to
  stdout, similar to other Cargo commands.

Changed

• In order to accommodate the machine-readable output, all of cargo-criterion's other output
  is now printed to stderr. This matches Cargo's normal behavior. If benchmark targets print
  anything to stdout, it will be redirected to stderr if --message-format is set, or will be
  left on stderr if not.
• Heavy internal refactoring of plot generation code. There may be some bugs.

1.0.0-alpha1

Added

• Initial version of cargo-criterion