As the title, what is exactly the difference of these two defs in Groovy?
Maybe it's a documentation problem, I can't find anything...
A method declaration without static marks a method as an instance method. Whereas a declaration with static will make this method static - can be called without creating an instance of that class - see https://www.geeksforgeeks.org/static-methods-vs-instance-methods-java/
def in groovy defines a value as duck typed. The capabilities of the value are not determined by its type, they are checked at runtime. The question if you can call a method on that value is answered at runtime - see optional typing.
static def means that the method will return a duck typed value and can be called without having instance of the class.
Example:
Suppose you have these two classes:
class StaticMethodClass {
static def test(def aValue) {
if (aValue) {
return 1
}
return "0"
}
}
class InstanceMethodClass {
def test(def aValue) {
if (aValue) {
return 1
}
return "0"
}
}
You are allowed to call StaticMethodClass.test("1"), but you have to create an instance of InstanceMethodClass before you can call test - like new InstanceMethodClass().test(true).
Related
Is there a way to validate a property value when the property is evaluated? I can't do it in the getter because that returns the Property object - I want the validation to run only when the actual value is calculated (i.e. I want to be lazy evaluation friendly).
They show extensions using the Property object here:
https://docs.gradle.org/current/userguide/lazy_configuration.html#connecting_properties_together
However, they don't explain how to do property validation when the value is calculated. Here is the snipet of code from the Gradle documentation provided example:
// A project extension
class MessageExtension {
// A configurable greeting
final Property<String> greeting
#javax.inject.Inject
MessageExtension(ObjectFactory objects) {
greeting = objects.property(String)
}
}
If I wanted to make sure the value of greeting was not equal to test, then how would I do that when it is evaluated?
For most use cases, it should be sufficient to just validate the property value once you resolve it in your task or in other internal parts of your plugin. Only a few extensions are actually designed to be consumed by other plugins or the build script.
Gradle does not provide some validation that can be attached to a property, however you can build this functionality on your own like in the example below:
class MessageExtension {
private final Property<String> _greeting
final Provider<String> greeting
#javax.inject.Inject
MessageExtension(ObjectFactory objects) {
_greeting = objects.property(String)
greeting = _greeting.map { value ->
if (value.equals('test'))
throw new RuntimeException('Invalid greeting')
return value
}
}
def setGreeting(String value) {
_greeting.set(value)
}
def setGreeting(Provider<String> value) {
_greeting.set(value)
}
}
project.extensions.create('message', MessageExtension)
message {
greeting = 'test'
}
println message.greeting.get()
I turned the Property into a backing field for a Provider that runs the validation when resolved. If you do not want to throw an exception, but just return an empty Provider, you may replace the map with a flatMap.
I would like to know by some form of reflection or other means if a given ES6 class has a user-written non-default constructor, or not.
Assuming that user-provided constructor has one argument or more, you can do that by checking the length property of the function(class). But if the constructor takes no argument, there is simply no way as far as I know
function Person(fName, lName) {
this.firstName = fName;
this.lastName = lName
}
console.log(Person.length);
function Person2() {}
console.log(Person2.length);
class Person3 {
constructor(f,l) {}
}
console.log(Person3.length);
class Person4 {
}
console.log(Person4.length);
You can invoke the Classname.prototype.constructor.toString() (where Classname is the inspected class name) and get the source string for the class. Which you can then parse and see if it was a constructor declared or not.
Presumably, you need a decent parser for that, but it's another story.
References:
http://www.ecma-international.org/ecma-262/6.0/#sec-function.prototype.tostring
I'm using a Supplier to instantiate a field thread safe while avoiding consecutive calls to the synchronized method.
class MyClass extends AbstractClassWithContext {
Supplier<Foo> fooGetter;
Foo foo;
public MyClass() {
this.fooGetter = this::initFoo;
}
Foo getFoo(){
return fooGetter.get();
}
synchonized Foo initFoo(){
if(Objects.isNull(this.foo)) {
this.foo = getContext().getFoo();
}
this.fooGetter = () -> this.foo;
return this.foo;
}
}
When I'm running my Unit Tests I want to make sure that initFoo() is called exactly once. Sadly verify(classUnderTest, times(1)).initFoo() does not register that initFoo is entered. I debugged this and calling getFoo() does in turn enter initFoo.
Any ideas?
I assume your test code looks something like this:
MyClass spiedOnObject = spy(new MyClass());
spiedOnObject.getFoo();
verify(spiedOnObject , times(1)).initFoo();
The problem is that this.fooGetter = this::initFoo; is called before you start spying on the object. At this point this refers to the real object, not to the spy. And that reference is captured when the method reference is created. Therefore the call cannot be registered.
We have a class like this in a Grails 2.4.3 application (migrated from 2.3.8):
#Validateable
class Foo {
Integer noDefault;
Integer withDefault = 1;
static constraints = {
noDefault(nullable:false)
withDefault(nullable:false)
}
}
This class is being instantiated in a complex configuration mechanism using a Map like this:
[
noDefault: 0,
withDefault: 2
]
(In fact the Map is part of a huge one, but the class constructor sees this small one.) Formerly the class worked if we omitted the withDefault entry from the config map, using the default value which is not null. In Grails 2.4.3, however, it tells me that this field cannot be null. I can fix it by letting it be null in the constraint, but it lets setting the explicite value null (and overwrite the default value), which causes problem during operation.
Do you know some workaround, which preserves the semantics and correct operation?
Thanx in advance, best regards: Balázs
What you are describing is not consistent with what I would expect and not consistent with the behavior I am seeing. The project at https://github.com/jeffbrown/validatedefaults contains the following code.
At https://github.com/jeffbrown/validatedefaults/blob/master/src/groovy/demo/Foo.groovy
// src/groovy/demo/Foo.groovy
package demo
import grails.validation.Validateable
#Validateable
class Foo {
Integer noDefault;
Integer withDefault = 1;
static constraints = {
noDefault(nullable:false)
withDefault(nullable:false)
}
}
The test at https://github.com/jeffbrown/validatedefaults/blob/master/test/unit/demo/FooSpec.groovy passes:
// test/unit/demo/FooSpec.groovy
package demo
import spock.lang.Specification
import grails.test.mixin.TestMixin
import grails.test.mixin.support.GrailsUnitTestMixin
#TestMixin(GrailsUnitTestMixin)
class FooSpec extends Specification {
void 'test validating default values'() {
given:
def map = [noDefault: 0]
def foo = new Foo(map)
expect:
foo.validate()
}
}
When I run the app I get the same behavior.
// grails-app/conf/BootStrap.groovy
import demo.Foo
class BootStrap {
def init = { servletContext ->
def map = [noDefault: 0]
def foo = new Foo(map)
// this prints true...
println "Foo is valid? : ${foo.validate()}"
}
def destroy = {
}
}
I hope that helps.
If not, is there anything like this on the horizon?
This is the one feature of JavaScript, Ruby, and Perl that I can't live without. I know you can fake it with a hash member, but I want to be able to create (arbitrary) "first class" members from a parser.
Currently there's nothing that can set a field that doesn't yet exist. The mirror API can be used to set fields that already exist, and may eventually be extended to support defining new fields dynamically.
You can also use the "noSuchMethod" method on a class to intercept setter / getter, and store the received value in a map.
For example (I can't remember the syntax exactly...):
class Foo {
var _dynamicProperties = new Map<String,Object>();
noSuchMethod(String function_name, List args) {
if (args.length == 0 && function_name.startsWith("get:")) {
// Synthetic getter
var property = function_name.replaceFirst("get:", "");
if (_dynamicProperties.containsKey(property)) {
return _dynamicProperties[property];
}
}
else if (args.length == 1 && function_name.startsWith("set:")) {
// Synthetic setter
var property = function_name.replaceFirst("set:", "");
// If the property doesn't exist, it will only be added
_dynamicProperties[property] = args[0];
return _dynamicProperties[property];
}
super.noSuchMethod(function_name, args)
}
}
And then you can use this in your code as follows:
var foo = new Foo();
foo.bar = "Hello, World!";
print(foo.bar);
Of course, this can lead to typos that will not be checked by the type checker, e.g.:
foo.bar = "Hello";
foo.baz = "Hello, World!"; // Typo, meant to update foo.bar.
There are ways you have type-checker validation by using redirecting factory constructors and an implied interface, but then it starts to get complicated.
Side note: This is what JsonObject uses to convert a JSON map to a class type syntax.