JavaFX Integration #4

Merged
kske merged 4 commits from f/javafx into develop 2021-12-16 12:05:50 +01:00
11 changed files with 328 additions and 6 deletions

View File

@ -1,11 +1,15 @@
package dev.kske.undoredo.core; package dev.kske.undoredo.core;
import java.util.List; import java.util.*;
/** /**
* A change manager keeps track of subsequent changes and allows un- and redoing them. A specific * A change manager keeps track of subsequent changes and allows un- and redoing them. A specific
* change can be marked using {@link #mark()} to keep track of a saved state in the application that * change can be marked using {@link #mark()} to keep track of a saved state in the application that
* uses the manager. * uses the manager.
* <p>
* If you intend to listen to the state of a change manager, consider writing a wrapper
* implementation for an existing change manager that adds the necessary hooks. If you use JavaFX,
* take a look at the {@code dev.kske.undoredo.javafx} module.
* *
* @param <C> the change type to store in this change manager * @param <C> the change type to store in this change manager
* @author Maximilian K&auml;fer * @author Maximilian K&auml;fer
@ -22,6 +26,12 @@ public interface ChangeManager<C extends Change> {
*/ */
void addChange(C change); void addChange(C change);
/**
* @return the change that was applied last
* @since 0.0.1
*/
Optional<C> getLastChange();
/** /**
* Undoes the current change. * Undoes the current change.
* *

View File

@ -22,6 +22,11 @@ public final class UnlimitedChangeManager<C extends Change> implements ChangeMan
++index; ++index;
} }
@Override
public Optional<C> getLastChange() {
return index < 0 ? Optional.empty() : Optional.of(changes.get(index));
kske marked this conversation as resolved Outdated
Outdated
Review

index < 0

`index < 0`
}
@Override @Override
public boolean undo() { public boolean undo() {
if (isUndoAvailable()) { if (isUndoAvailable()) {

View File

@ -27,20 +27,33 @@ class ChangeManagerTest {
* @since 0.0.1 * @since 0.0.1
*/ */
@Test @Test
@Order(1) @Order(10)
void testAddChange() { void testAddChange() {
assertSame(0, wrapper.value); assertSame(0, wrapper.value);
manager.addChange(change); manager.addChange(change);
assertSame(1, wrapper.value); assertSame(1, wrapper.value);
} }
/**
* Tests retrieving the last change.
*
* @since 0.0.1
*/
@Test
@Order(20)
void testLastChange() {
assertTrue(manager.getLastChange().isEmpty());
manager.addChange(change);
assertEquals(change, manager.getLastChange().get());
}
/** /**
* Tests the consistency of the change list. * Tests the consistency of the change list.
* *
* @since 0.0.1 * @since 0.0.1
*/ */
@Test @Test
@Order(2) @Order(30)
kske marked this conversation as resolved Outdated
Outdated
Review

I see you noticed the problem when adding a new test that has to be executed earlier:
You have to increment all subsequent priorities.
For this reason I propose another mechanism for using @Order: Instead of @Order(n), we should use @Order(10^n) for the tests that are created when creating the class.
Adding newer tests is then as simple as using @Order(10^n / i) where i is the amount of other elements with that order.
That way, when we have the basic tests, we don't have to subsequently change all indices below them when adding newer tests.

Take the following example:
@Order(1), @Order(2), @Order(3), @Order(4), @Order(5) are the tests we created initially.
Now, we add @Order(2b), @Order(3a), @Order(3c) and @Order(6). With our mechanism, that is pretty inefficient.
With my method, we would have initially
@Order(10), @Order(100), @Order(1 000), @Order(10 000), @Order(100 000).
Now, we simply add @Order(150), @Order(500), @Order(1 500) and @Order(1 000 000).
And then we're done.

I see you noticed the problem when adding a new test that has to be executed earlier: You have to increment all subsequent priorities. For this reason I propose another mechanism for using `@Order`: Instead of `@Order(n)`, we should use `@Order(10^n)` for the tests that are created when creating the class. Adding newer tests is then as simple as using `@Order(10^n / i)` where `i` is the amount of other elements with that order. That way, when we have the basic tests, we don't have to subsequently change all indices below them when adding newer tests. Take the following example: `@Order(1)`, `@Order(2)`, `@Order(3)`, `@Order(4)`, `@Order(5)` are the tests we created initially. Now, we add `@Order(2b)`, `@Order(3a)`, `@Order(3c)` and `@Order(6)`. With our mechanism, that is pretty inefficient. With my method, we would have initially `@Order(10)`, `@Order(100)`, `@Order(1 000)`, `@Order(10 000)`, `@Order(100 000)`. Now, we simply add `@Order(150)`, `@Order(500)`, `@Order(1 500)` and `@Order(1 000 000)`. And then we're done.
Outdated
Review

How would you deal with a class that has more than nine tests?

How would you deal with a class that has more than nine tests?
Outdated
Review

Bring out my "division by 2" skills again.
Also, you first need to have 9 tests that depend on order.
In most cases, tests should not depend on a specific order.

Bring out my "division by 2" skills again. Also, you first need to have 9 tests that depend on order. In most cases, tests should not depend on a specific order.
Outdated
Review

I would prefer having gaps of 10, or maybe 100, but with exponential gaps, the rule is unnecessarily complex and we can easily surpass Integer.MAX_VALUE / 2, which is the default order for JUnit tests, leading to unexpected results for tests without an explicitly defined order.

I would prefer having gaps of 10, or maybe 100, but with exponential gaps, the rule is unnecessarily complex and we can easily surpass `Integer.MAX_VALUE / 2`, which is the default order for JUnit tests, leading to unexpected results for tests without an explicitly defined order.
Outdated
Review

That's another possibility.
I don't care.
Should we use the notation @Order(10n) then?

That's another possibility. I don't care. Should we use the notation `@Order(10n)` then?
void testGetChanges() { void testGetChanges() {
assertTrue(manager.getChanges().isEmpty()); assertTrue(manager.getChanges().isEmpty());
manager.addChange(change); manager.addChange(change);
@ -54,7 +67,7 @@ class ChangeManagerTest {
* @since 0.0.1 * @since 0.0.1
*/ */
@Test @Test
@Order(2) @Order(40)
void testUndo() { void testUndo() {
assertFalse(manager.isUndoAvailable()); assertFalse(manager.isUndoAvailable());
assertFalse(manager.undo()); assertFalse(manager.undo());
@ -63,6 +76,7 @@ class ChangeManagerTest {
assertTrue(manager.undo()); assertTrue(manager.undo());
assertFalse(manager.isUndoAvailable()); assertFalse(manager.isUndoAvailable());
assertFalse(manager.undo()); assertFalse(manager.undo());
assertTrue(manager.getLastChange().isEmpty());
} }
/** /**
@ -71,7 +85,7 @@ class ChangeManagerTest {
* @since 0.0.1 * @since 0.0.1
*/ */
@Test @Test
@Order(4) @Order(50)
void testRedo() { void testRedo() {
assertFalse(manager.isRedoAvailable()); assertFalse(manager.isRedoAvailable());
assertFalse(manager.redo()); assertFalse(manager.redo());
@ -83,6 +97,7 @@ class ChangeManagerTest {
assertTrue(manager.redo()); assertTrue(manager.redo());
assertFalse(manager.isRedoAvailable()); assertFalse(manager.isRedoAvailable());
assertFalse(manager.redo()); assertFalse(manager.redo());
assertEquals(change, manager.getLastChange().get());
} }
/** /**
@ -91,7 +106,7 @@ class ChangeManagerTest {
* @since 0.0.1 * @since 0.0.1
*/ */
@Test @Test
@Order(5) @Order(60)
void testMark() { void testMark() {
assertTrue(manager.isAtMarkedIndex()); assertTrue(manager.isAtMarkedIndex());
manager.addChange(change); manager.addChange(change);

27
javafx/.classpath Normal file
View File

@ -0,0 +1,27 @@
<?xml version="1.0" encoding="UTF-8"?>
<classpath>
<classpathentry kind="src" output="target/classes" path="src/main/java">
<attributes>
<attribute name="optional" value="true"/>
<attribute name="maven.pomderived" value="true"/>
</attributes>
</classpathentry>
<classpathentry kind="src" output="target/test-classes" path="src/test/java">
<attributes>
<attribute name="optional" value="true"/>
<attribute name="maven.pomderived" value="true"/>
<attribute name="test" value="true"/>
</attributes>
</classpathentry>
<classpathentry kind="con" path="org.eclipse.jdt.launching.JRE_CONTAINER/org.eclipse.jdt.internal.debug.ui.launcher.StandardVMType/JavaSE-11">
<attributes>
<attribute name="maven.pomderived" value="true"/>
</attributes>
</classpathentry>
<classpathentry kind="con" path="org.eclipse.m2e.MAVEN2_CLASSPATH_CONTAINER">
<attributes>
<attribute name="maven.pomderived" value="true"/>
</attributes>
</classpathentry>
<classpathentry kind="output" path="target/classes"/>
</classpath>

23
javafx/.project Normal file
View File

@ -0,0 +1,23 @@
<?xml version="1.0" encoding="UTF-8"?>
<projectDescription>
<name>undo-redo-javafx</name>
<comment></comment>
<projects>
</projects>
<buildSpec>
<buildCommand>
<name>org.eclipse.jdt.core.javabuilder</name>
<arguments>
</arguments>
</buildCommand>
<buildCommand>
<name>org.eclipse.m2e.core.maven2Builder</name>
<arguments>
</arguments>
</buildCommand>
</buildSpec>
<natures>
<nature>org.eclipse.jdt.core.javanature</nature>
<nature>org.eclipse.m2e.core.maven2Nature</nature>
</natures>
</projectDescription>

28
javafx/pom.xml Normal file
View File

@ -0,0 +1,28 @@
<project xmlns="http://maven.apache.org/POM/4.0.0"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 https://maven.apache.org/xsd/maven-4.0.0.xsd">
<modelVersion>4.0.0</modelVersion>
<artifactId>javafx</artifactId>
<name>Undo-Redo JavaFX Integration</name>
<parent>
<groupId>dev.kske</groupId>
<artifactId>undo-redo</artifactId>
<version>0.0.1-SNAPSHOT</version>
</parent>
<dependencies>
<dependency>
<groupId>dev.kske</groupId>
<artifactId>undo-redo-core</artifactId>
<version>0.0.1-SNAPSHOT</version>
</dependency>
<dependency>
<groupId>org.openjfx</groupId>
<artifactId>javafx-base</artifactId>
<version>11</version>
Review

Shouldn't it be provided?

Otherwise we run into version problems...

Shouldn't it be `provided`? Otherwise we run into version problems...
Review

A project using this library should be able to use any JavaFX version >= 11. A fixed version would be defined like this: [11]. <scope>provided</scope> assumes that the dependency is present in some kind of runtime container, like the JDK or a Java EE container, which isn't the case here.

A project using this library should be able to use any JavaFX version >= 11. A fixed version would be defined like this: `[11]`. `<scope>provided</scope>` assumes that the dependency is present in some kind of runtime container, like the JDK or a Java EE container, which isn't the case here.
Review

I thought that was exactly the goal behind provided: To declare that any version of JavaFX is possible, as long as it is present...

I thought that was exactly the goal behind `provided`: To declare that any version of JavaFX is possible, as long as it is present...
Review

I don't think the scope has anything to do with the version: https://maven.apache.org/guides/introduction/introduction-to-dependency-mechanism.html#Dependency_Scope

There is a possibility to specify version ranger, but the default should be sufficient for us: https://docs.oracle.com/middleware/1212/core/MAVEN/maven_version.htm#MAVEN402

I don't think the scope has anything to do with the version: https://maven.apache.org/guides/introduction/introduction-to-dependency-mechanism.html#Dependency_Scope There is a possibility to specify version ranger, but the default should be sufficient for us: https://docs.oracle.com/middleware/1212/core/MAVEN/maven_version.htm#MAVEN402
</dependency>
</dependencies>
</project>

View File

@ -0,0 +1,134 @@
package dev.kske.undoredo.javafx;
import java.util.List;
import javafx.beans.property.*;
import dev.kske.undoredo.core.*;
/**
* Wraps an ordinary change manager into an observable change manager, providing the required
* properties for concrete implementations.
kske marked this conversation as resolved
Review
<p>
The properties have exactly the same name as their corresponding `-property()` methods and can be called i.e. as binding under exactly that name. Alternatively, the names are available as constants.
```java <p> The properties have exactly the same name as their corresponding `-property()` methods and can be called i.e. as binding under exactly that name. Alternatively, the names are available as constants. ```
* <p>
* The properties have the same name as their corresponding {@code -property()} methods and can be
* accessed reflectively from JavaFX, e.g. through
* {@link javafx.beans.binding.Bindings#select(Object, String...)}. Alternatively, the property
* names are available as constants.
*
* @param <C> the change type to store in this change manager
* @param <M> the type of change manager to wrap
* @author Kai S. K. Engelbart
* @since 0.0.1
Review

I have to say, I like that you were even able to implement it without having to use listeners.
Maybe we should add a paragraph in the Javadoc of ChangeManager that that is the preferred way to listen to changes.

I also like that our implementation seems to be

  1. less filesystem-space consuming (amount of classes and modules)
  2. less memory consuming during runtime (we only use the necessary objects unlike UndoFX which creates an object storing the current position whenever you undo or redo something...)
  3. far more readable (I have no idea what UndoFX is doing at any point)
  4. without weird dependencies that make no sense at all
  5. conformant with JavaFX coding standard
  6. more performant as there are no listeners and callbacks and equals that are always tested
I have to say, I like that you were even able to implement it without having to use listeners. Maybe we should add a paragraph in the Javadoc of `ChangeManager` that that is the preferred way to listen to changes. I also like that our implementation seems to be 1. less filesystem-space consuming (amount of classes and modules) 2. less memory consuming during runtime (we only use the necessary objects unlike `UndoFX` which creates an object storing the current position whenever you undo or redo something...) 3. far more readable (I have no idea what UndoFX is doing at any point) 4. without weird dependencies that make no sense at all 5. conformant with JavaFX coding standard 6. more performant as there are no listeners and callbacks and equals that are always tested
Review

UndoFX might be more flexible in some cases (e.g. non-linear history and such), but for our and most other purposes this solution should be more than enough.

UndoFX might be more flexible in some cases (e.g. non-linear history and such), but for our and most other purposes this solution should be more than enough.
*/
kske marked this conversation as resolved
Review

Perhaps we should extract these names into public static final constants to avoid confusion.

Perhaps we should extract these names into `public static final` constants to avoid confusion.
Review

You mean the string literal? So for every property that is declared we would declare another constant?

You mean the string literal? So for every property that is declared we would declare another constant?
Review

Yes.

Yes.
Review

But what purpose should this have? When would it be used?

But what purpose should this have? When would it be used?
Review

Bindings.bind(someProperty, ChangeManagerWrapper.UndoAvailable) or however it is called.

`Bindings.bind(someProperty, ChangeManagerWrapper.UndoAvailable)` or however it is called.
public class ChangeManagerWrapper<C extends Change, M extends ChangeManager<C>>
implements ObservableChangeManager<C> {
public static final String LAST_CHANGE = "lastChange";
public static final String AT_MARKED_INDEX = "atMarkedIndex";
public static final String UNDO_AVAILABLE = "undoAvailable";
public static final String REDO_AVAILABLE = "redoAvailable";
protected ReadOnlyObjectWrapper<C> lastChange =
new ReadOnlyObjectWrapper<>(this, LAST_CHANGE);
kske marked this conversation as resolved Outdated
Outdated
Review

Wait, protected?
How do you instantiate one then? Do you really have to create your own class? Isn't that a bit overkill?

Wait, protected? How do you instantiate one then? Do you really have to create your own class? Isn't that a bit overkill?
Outdated
Review

I think the wrapper can be concrete, but should be open for extension in case we want a wrapper for a change manager with additional methods.

I think the wrapper can be concrete, but should be open for extension in case we want a wrapper for a change manager with additional methods.
protected ReadOnlyBooleanWrapper atMarkedIndex =
new ReadOnlyBooleanWrapper(this, AT_MARKED_INDEX);
protected ReadOnlyBooleanWrapper undoAvailable =
new ReadOnlyBooleanWrapper(this, UNDO_AVAILABLE);
protected ReadOnlyBooleanWrapper redoAvailable =
new ReadOnlyBooleanWrapper(this, REDO_AVAILABLE);
protected final M manager;
/**
* Initializes a change manager wrapper.
*
* @param manager the change manager to wrap
* @since 0.0.1
*/
public ChangeManagerWrapper(M manager) {
this.manager = manager;
}
@Override
public void addChange(C change) {
manager.addChange(change);
updateProperties();
}
delvh marked this conversation as resolved
Review

Wait, what was the result of the last discussion we had about that?
I can't remember, and I don't remember either which PR in which project it was where we talked about that.

Wait, what was the result of the last discussion we had about that? I can't remember, and I don't remember either which PR in which project it was where we talked about that.
Review

We discussed this in #2. I think @DieGurke resolved the suggestion as we didn't reach a conclusion after stating our opinions.

We discussed this in #2. I think @DieGurke resolved the suggestion as we didn't reach a conclusion after stating our opinions.
@Override
public boolean undo() {
if (manager.undo()) {
updateProperties();
return true;
}
return false;
}
@Override
public boolean redo() {
if (manager.redo()) {
updateProperties();
return true;
}
return false;
}
@Override
public void mark() {
manager.mark();
setAtMarkedIndex(manager.isAtMarkedIndex());
}
/**
* Sets the values of all properties to those present in the wrapped change manager.
*
* @since 0.0.1
*/
private void updateProperties() {
setLastChange(manager.getLastChange().orElse(null));
setAtMarkedIndex(manager.isAtMarkedIndex());
setUndoAvailable(manager.isUndoAvailable());
setRedoAvailable(manager.isRedoAvailable());
}
@Override
public final ReadOnlyObjectProperty<C> lastChangeProperty() {
return lastChange.getReadOnlyProperty();
}
protected final void setLastChange(C lastChange) {
this.lastChange.set(lastChange);
}
@Override
public final ReadOnlyBooleanProperty atMarkedIndexProperty() {
return atMarkedIndex.getReadOnlyProperty();
}
protected final void setAtMarkedIndex(boolean atMarkedIndex) {
this.atMarkedIndex.set(atMarkedIndex);
}
@Override
public final ReadOnlyBooleanProperty undoAvailableProperty() {
return undoAvailable.getReadOnlyProperty();
}
protected final void setUndoAvailable(boolean undoAvailable) {
this.undoAvailable.set(undoAvailable);
}
@Override
public final ReadOnlyBooleanProperty redoAvailableProperty() {
return redoAvailable.getReadOnlyProperty();
}
protected final void setRedoAvailable(boolean redoAvailable) {
this.redoAvailable.set(redoAvailable);
}
@Override
public List<C> getChanges() {
return manager.getChanges();
}
}

View File

@ -0,0 +1,47 @@
package dev.kske.undoredo.javafx;
import java.util.Optional;
import javafx.beans.property.*;
import dev.kske.undoredo.core.*;
/**
* A change manager that exposes its state through JavaFX properties, thereby allowing a direct
kske marked this conversation as resolved Outdated
Outdated
Review

Purpose of this interface?

Purpose of this interface?
* integration of Undo-Redo with JavaFX listeners and property bindings.
*
* @param <C> the change type to store in this change manager
* @author Kai S. K. Engelbart
* @since 0.0.1
* @see ChangeManagerWrapper
*/
public interface ObservableChangeManager<C extends Change> extends ChangeManager<C> {
ReadOnlyObjectProperty<C> lastChangeProperty();
@Override
default Optional<C> getLastChange() {
return Optional.of(lastChangeProperty().get());
}
ReadOnlyBooleanProperty atMarkedIndexProperty();
@Override
default boolean isAtMarkedIndex() {
return atMarkedIndexProperty().get();
}
ReadOnlyBooleanProperty undoAvailableProperty();
@Override
default boolean isUndoAvailable() {
return undoAvailableProperty().get();
}
ReadOnlyBooleanProperty redoAvailableProperty();
@Override
default boolean isRedoAvailable() {
return redoAvailableProperty().get();
}
}

View File

@ -0,0 +1,7 @@
/**
* Contains JavaFX-based wrapper API for integrating Undo-Redo with JavaFX.
*
* @author Kai S. K. Engelbart
* @since 0.0.1
*/
package dev.kske.undoredo.javafx;

View File

@ -0,0 +1,15 @@
/**
* Contains JavaFX-based wrapper API for integrating Undo-Redo with JavaFX.
*
* @author Kai S. K. Engelbart
* @since 0.0.1
*/
module dev.kske.undoredo.javafx {
exports dev.kske.undoredo.javafx;
opens dev.kske.undoredo.javafx to javafx.base;
kske marked this conversation as resolved Outdated
Outdated
Review

requires transitive?

`requires transitive`?
Outdated
Review

The core module doesn't have any dependencies, so this shouldn't make a difference.

The core module doesn't have any dependencies, so this shouldn't make a difference.
Outdated
Review

I think transitive is meant for the other way around:
Not that we require all dependencies from that module,
but that modules requiring our module also automatically require the core module.

I think `transitive` is meant for the other way around: Not that we require all dependencies from that module, but that modules requiring our module also automatically require the `core` module.
requires transitive dev.kske.undoredo.core;
requires transitive javafx.base;
}

11
pom.xml
View File

@ -14,6 +14,7 @@
<modules> <modules>
<module>core</module> <module>core</module>
<module>javafx</module>
</modules> </modules>
<licenses> <licenses>
@ -47,6 +48,16 @@
</roles> </roles>
<timezone>Europe/Berlin</timezone> <timezone>Europe/Berlin</timezone>
</developer> </developer>
<developer>
<name>Leon Hofmeister</name>
<email>leon@kske.dev</email>
<url>https://git.kske.dev/delvh</url>
<roles>
<role>architect</role>
<role>developer</role>
</roles>
<timezone>Europe/Berlin</timezone>
</developer>
</developers> </developers>
<scm> <scm>