v6.7.0

6.7.0

Upgrade Instructions

• It is not required, but you should remove the existing Gut node for any scenes you have that use it and then re-add it and re-configure it. Re-adding will get rid of the caution symbol next to the control (this is due to changes in inheritance, Gut changed from a WindowDialog to a Control)
• For the command line, note that the log option in the .gutconfig.json file has changed to log_level for consistency.

Features

• 3.1 Compatible (with a few very minor issues, check them out here.)
• Issue 73 You can now export your tests and run them in your exported game! This means you can run your tests on all the platforms that Godot supports through the executable...no editor, no nothing! Check out the Export Tests wiki page for more information.
• Issue 102 Added get_call_parameters which allows you to get the parameters sent to a call to a doubled method. It returns the most recent call by default but you can specify an optional index as well.
• Issue 103 Added replace_node which allows you to replace a node in a scene with another node by specifying the node path. This makes testing scenes that use the $ syntax to get a reference to a child node easier to test by giving you a quick way to inject a double of any child node. The node that was replaced is freed with queue_free
• A brand new GUI. It looks a lot like the old GUI but it is new. With the ability to export tests I wanted the GUI to be more mobile friendly. So I threw the old one out (which was a relic of the days when Gut was one script...the old GUI was made entirely with code) and created a scene which will make GUI changes soooo much easier going forward.
• -gprint_gutconfig_sample command line option will print you a full config file that you can start using.

Fixes/Improvements

• Issue 104 Added all missing settings to the gutconfig file.
  • For consistency the log option in the .gutconfig changed to log_level.
• Improved logging. More warnings and errors have been added to help when something goes wrong inside Gut. The number of Gut related errors, warnings, and deprecated methods are listed in the summary. If you don't see any listed, there aren't any.
• A lot of housekeeping and Boy Scouting.

Removals

• You can no longer set the selected script when adding scripts one by one via add_script. If you pass a 2nd parameter an error is generated, but everything will continue to run.
• Deprecated the following methods:
  • end_test and the old gut.end_yielded_test. These no longer need to be called.
  • assert_get_set_methods was replaced with assert_accessors awhile ago. In this release I added deprecation messages for this method. It still works, but wanted to start getting the message out.
  • assert_extends has been replaced with assert_is to match the 3.0 names.
• Stop button was removed. It didn't really work that well anyhow. I could be convinced to put it back.

v6.6.3

Missed a new directory in the .gitattributes.

6.6.2

6.6.2

Features

• assert_null and assert_not_null
• added simulate to the test object so you no longer have to use gut when calling it. It is fine if you still do though. Sorry if this breaks something.
• Doubling, Stubbing, and Spies are no longer considered experimental.
• The double method is now smart:
  • It knows when you pass it a scene or a script, no need for double_scene anymore but it remains.
  • You can pass it a loaded class or scene like this:

  var MyClass = load('res://my_class.gd')
  var MyClassDouble = double(MyClass)
  

• You can now double Inner Classes by passing an Inner Class path like so:

  double('res://my_script.gd', 'Inner1/InnerInInner1/AndSoOn')
  

• The start of an internal Gut logger to make better messages down the road.
• Experimental Full doubling allows you to Spy on most Built-in methods. You still cannot Stub them though. This must be enabled, details in the Double Wiki page.
• Added link to Rainware's setup tutorial on youtube to the README

Fixes

• Issue 94 Gut runs after_all on inner classes that are skipped.
• Rawsyntax fixed a bunch of misspellings so that we can erock on with less bad speligs fo wrds.

6.6.1

Major bug:

• Issue 95: Fixed issue where sometimes Gut can end up clearing files from res:\\ when using doubling.

Enhancements

• Issue 60: Improved signal assert failure messages in some cases by having them include a list of signals that were emitted for the object.

Minor bug

• Issue 88: Fixed issue where you couldn't specify the config_file from the command line.

6.6.0

6.6.0

Fixes

• Issue 79: Scaling wasn't being accounted for by the maximize method.
• Issue 80: Inner classes are no longer included in the count of scripts that were ran.

Features

Issue 83: Added option to include subdirectories. Thanks to ChemicalInk for doing the initial work to implement the traversals. The option is off by default so that it doesn't break anything. This will probably change in later releases.

• gutconfig: include_subdirs
• command line: -ginclude_subdirs
• editor: Include Subdirectories
• Gut instance: set/get_include_subdirectories()

Issue 69: Renamed some methods. The old names will remain but the documentation has been updated to reflect the new names. If I ever end up removing the old names they will become deprecated for awhile and then removed in some later release.

• assert_get_set_methods renamed to assert_accessors b/c it is easier to say
• assert_extends renamed to assert_is b/c the keyword changed in gut 3.0

The setup/teardown methods got a rename so they are a little easier to understand. You should use the new names going forward, but refactoring existing tests can be messy (see note below).

• prerun_setup renamed to before_all
• setup renamed to before_each
• postrun_teardown renamed to after_all
• teardown renamed to after_each

Note about setup/teardown rename:

• The new methods could not be actual aliases due to how they are used internally. They exist side by side with the old names and are called after the old methods. DO NOT use both in the same script.
• If you refactor your tests to use the new names, be careful wherever you have a test class that extends another test class and it calls super's setup/teardown methods. For example, if you rename super's setup to before_each but all the test classes that inherit from it are still calling .setup then you'll have problem on your hands.

6.5.0

Fixes

• Bug fix by Xrayez to ensure that the command line tool always sets the return code properly. Before it was only setting it if Gut was configured to exit when done.
• Fixed an issue where the command line tool wasn't setting options correctly when the .gutconfig.json file was present. All options are now applied correctly based on order of precedence (default < gutconfig < cmd line). I also added the -gpo command line option to print out all option values from all sources and what value would be used when running Gut. This will make debugging theses issues easier later.

Features

• We have two new asserts thanks to hbergren. These asserts make it easier to assert if a value is within or outside of a +/- range of a value. These are especially useful when comparing floats that the engine insists aren't equal due to rounding errors.
  • assert_almost_eq(got, expected, error_interval, text='') - Asserts that got is within the range of expected +/- error_interval. The upper and lower bounds are included in the check. Verified to work with integers, floats, and Vector2. Should work with anything that can be added/subtracted. Examples
  • assert_almost_ne(got, expected, error_interval, text='') - This is the inverse of assert_almost_eq. This will pass if got is outside the range of expected +/- error_interval. Examples
• Xrayez contributed a new option to maximize the Gut window upon launch. The option can be set in the editor, .gutconfig, or at the command line.
• Added the -gpo command line option to print out all option values from all sources and what value would be used when running Gut. This will make debugging option issues much easier.

Other

Some housekeeping. Removed some commented out and unreachable code. Renamed a lot of tests in test_test.gd since it now uses Inner Test Classes which allows for better names. They were setting a bad example for PRs.

Spies and some other wonders.

I've "dog food"ed the doubles, stubs, and spies more in my own game and I think they are pretty stable. This release contains some tweaks to doubles and stubs and the introduction of spies as well as some other testing goodness.

Features

• inner_class_name option for editor node, command line, and .gutconfig.json.
• assert_exports: Assert that script exports a variable with a specific type. Thanks Myrkrheim
• Command line now returns 0 when all tests pass and 1 if there are any failures. Thanks cmfcmf.
• You can now set the opacity for the GUI through a command line option or the .gutconfig.json file. That one is also thanks to Myrkheim.
• Spies (experimental)
  • You can make assertions now about method calls on doubled objects.
  • assert_called
  • assert_not_called
  • assert_call_count

Fixes

• Fixed issue with duplicate methods in doubled classes.

Inner Test classes, doubles, stubs, and more.

6.3.0

Wiki

First, the Readme has been broken up into a Github wiki. I think this will make consuming the help easier. There is probably some room for improvement in the documentation and organization. Create issues for anything that could be done better.

Inner Test Classes

You can now create "Inner Classes" that run tests. This allows you to breakup your tests and create multiple setup/teardown scopes in a single script.

Example:

extends "res://addons/gut/test.gd"

class TestInnerClass:
  extends "res://addons/gut/test.gd"

  func setup():
    pass

class TestInnerClass2:
  extends "res://addons/gut/test.gd"

  func setup():
    pass

Command line config file

There is now support for a config file for the command line tool. It only supports some options right now, but that will be expanded in the future.

{
  "dirs":["res://test/unit/", "res://test/integration/"],
  "should_exit":true,
  "ignore_pause":true,
  "log": 3
}

Experimental Doubles and Stubs

There are also two new experimental features in this release. Doubling and Stubbing. These are experimental because their implementation could change a lot. I hope not, but they might. I fought with how these should work a lot, and honestly, I might have gotten it wrong. So I'm going to put it out there and see how they get used in the wild. I'm dedicated to keeping these features and improving on them, but they might be a little rough around the edges.

You can find out more about Stubbing and Doubling in the Wiki.

Some new asserts!

6.2.0

• Some new asserts courtesy of Myrkrheim

assert_string_contains
assert_string_starts_with
assert_string_ends_with
assert_has_method

• Added .gitattributes which might make for less un-checkboxing when installing from Asset Library and less extra files when downloading the addon. If this works out, then the next release will undo moving everything into gut_tests_and_examples and everything will be right in the world. Thanks cmfcmf.

6.1.0

• Moved as many files as I could to gut_tests_and_examples so that there was less stuff to uncheck when installing via the in-engine Asset Library. I'm still not 100% happy with the setup.
• Moved the License to addons/gut/ so that it is distributed with the addon and doesn't accidently get copied into the root of some other project when installed via the Asset Library.
• Some README tweaks.
• Fixed resize window handle bug. It was connecting to wrong signals and didn't work.
• Missed changing simulate to call _physics_process instead of _fixed_process in the 3.0 conversion. Fixed that.
• Improved summary. It now lists all failures and pendings instead of just listing the scripts that have failures or pending tests.
• Fixed issue where the signal_watcher could try to disconnect from a freed object.
• Added yield_to which allows you to yield to a signal or a maximum amount of time. This keeps your tests moving along if you yield to a signal that never gets emitted. Now the test will fail after an amount of time instead of sitting in limbo forever. This will also watch the signals on the object so you can make asserts about signals after the yield and you can save a line of code.
  Example:

# wait for my_object to emit the signal 'my_signal'
# or 5 seconds, whichever comes first.
yield(yield_to(my_object, 'my_signal', 5), YIELD)  
assert_signal_emitted(my_object, 'my_signal', \
                     'Maybe it did, maybe it didnt, but we still got here.')