1.3.2

Added "Reverse colors" option to visually swap the coloring so no need to reconfigure the coloring just because in a benchmark, smaller means better (green):

Column delete bugfix

Fixes a small crash when the comparison was active and a column was deleted.

Screenshot, icons, percentages

This release has three notable improvements:

• Take a screenshot of the table and put it onto the clipboard; no more manual print-screen and cutting the image around manually.
• Management buttons now have icons instead of text; reduces the spanning of the toolbar and leaves room for further buttons.
• Display the percentage difference relative to the baseline; comes in handy when quantifying improvements or regressions besides the high-level coloring.

Display errors, GUI improvements

Notable changes:

• Column-related operations have been moved into a popup menu:

• Tabs and columns can be easily duplicated now.
• The column or tab name no longer turns into null if the user cancels the input dialog.
• Added the ability to remove selection and focus from the table which makes it easier to take screenshots of it.
• Fixed empty parameter values if the pivoted table didn't have data for every column in certain rows.
• Added the ability to display the errors besides the measured values. They receive the same coloring as the measurement value when comparing against a baseline. This works with existing data.

v1.1.0: Export, pivot and paste enhancements

Bugfixes

• Fixed color options not saved properly on exit.

Export

The currently selected tab's contents can now be exported into a CSV or "XLS" format. Experience shows that CSV parsing is highly locale dependent, especially with Excel, therefore, the File > Export settings... option let's you specify the CSV column separator (usually comma or semicolon) and whether the floating point numbers should use dot or the current locale-specific decimal separator (i.e., comma).

The "XLS" format is not really a binary Excel file but rather a HTML file containing a single table (with colors if a comparison baseline column was selected). Even though some Excel versions will complain when opening the generated file, it should display properly.

Paste and pivot

This is a new feature that allows using one of the parameter columns of the JMH results as pivot columns, and this column's distinct values become the columns of the table.

For examle, if you benchmarked some code with a parameter that enables or disables certain behavior, you can pivot based on this parameter and compare the runs side-by-side.

Benchmark                                       (times)   Mode   Samples        Score  Score error    Units
r.i.AtomicPerf.atomicIntCASCheckFailure               1   avgt         5        2,668        0,057    ns/op
r.i.AtomicPerf.atomicIntCASCheckFailure            1000   avgt         5     1534,512       74,719    ns/op
r.i.AtomicPerf.atomicIntCASCheckFailure         1000000   avgt         5  2284937,183    28254,968    ns/op
r.i.AtomicPerf.atomicIntCASCheckSuccess               1   avgt         5       16,064        0,576    ns/op
r.i.AtomicPerf.atomicIntCASCheckSuccess            1000   avgt         5    11408,954      213,337    ns/op
r.i.AtomicPerf.atomicIntCASCheckSuccess         1000000   avgt         5 11417083,239   142717,458    ns/op

Note that this option clears the contents of the current tab.

v1.0.5: More forgiving JMH output parsing

This release contains parsing changes to accept more variety of JMH output formats, i.e., surrounded by spaces, embedded in comment sections or containing empty lines between values.

Related issues:

• #1: skip empty lines
• #4: trim leading/trailing spaces and comment symbols
• #5: ignore additional text between score and error among the value lines.