How do I fix this "Constructor cannot be called on object type" error in flow? - flowtype

I'm having trouble figuring out the problem that flow is complaining about. I'm trying to allow the implementation of an API be changeable by storing the implementation class, then later instantiating it, however, flow complains when I call new this.implKlass saying that "Constructor cannot be called on object type". What is flow trying to tell me, and what am I conceptually missing about how flow works?
Example code below, and flow try code here
/* #flow */
type ApiT = {
fnA(): Promise<*>;
}
// An implementation of the API
class Impl {
async fnA(): Promise<*> { return 1; }
}
class DoThings {
implKlass: ApiT;
constructor(klass) {
this.implKlass = klass;
}
callA() {
const Klass = this.implKlass;
const inst = new Klass();
return inst.fnA();
}
}
new DoThings(Impl).callA();
Example output:
18: const inst = new Klass();
^ constructor call. Constructor cannot be called on
18: const inst = new Klass();
^ object type
13: constructor(klass: ApiT) {
^ property `fnA`. Property not found in
23: new DoThings(Impl).callA();
^ statics of Impl

With a small modification this works.
class DoThings {
implKlass: Class<ApiT>;
constructor(klass) {
this.implKlass = klass;
}
callA() {
const Klass = this.implKlass;
const inst = new Klass();
return inst.fnA();
}
}
The bug was you were writing ApiT instead of Class<ApiT>. ApiT would be an instance of a class, while Class<ApiT> is the class itself.
Try flow link

ApiT describes an object type, not a class type. An instance of the Impl class satisfies the ApiT type, but the class Impl itself does not. You cannot call Impl.fnA(), for example.
I'm not sure if there is any way to pass around constructors like this. However you can accomplish basically the same thing by using a factory function:
type ApiT = {
fnA(): Promise<*>;
}
type ApiTFactory = () => ApiT;
class Impl {
async fnA(): Promise<*> { return 1; }
}
class DoThings {
factory: ApiTFactory;
constructor(factory: ApiTFactory) {
this.factory = factory;
}
callA() {
const factory = this.factory;
const inst = factory();
return inst.fnA();
}
}
new DoThings(() => new Impl()).callA();
tryflow link

Related

TypeScript Compiler API: How to get type with resolved type arguments?

I want to merge class declarations in a .dt.s file to generate a cleaner public API. I am stuck on how to make this work with generic type arguments. Let's say I have:
class A1<T> { // Non-exported class I want to hide
data?: T;
}
export class B1 extends A1<string> {
}
Ideally, I want to turn this into:
export class B1 {
data?: string;
}
I can get the type of A1 and then copy its members. But how do get a resolved version of A1 that uses string instead of T?
For reference, this is my current code:
for (const heritageClause of node.heritageClauses) {
for (const type of heritageClause.types) {
if (isExported(type.modifiers)) {
exportedTypes.push(type);
} else {
const privateType = typeChecker.getTypeAtLocation(type);
if (privateType?.symbol?.members) {
privateType.symbol.members.forEach((definition, memberName) => {
if (!currentMembers || !currentMembers.has(memberName)) {
additionalMembers.push(...definition.declarations);
}
}
});
}
}
}
I believe the method you are looking for is TypeChecker#getTypeOfSymbolAtLocation(symbol, node).
The following should get the resolved type of string | undefined:
// by the way, recommend renaming `type` to `typeNode` to avoid confusion
typeChecker.getTypeOfSymbolAtLocation(privateType.getProperties()[0], type);

Haxe: Binding pattern with abstract fields access methods

I'd like to make wrapper to implement simple data binding pattern -- while some data have been modified all registered handlers are got notified. I have started with this (for js target):
class Main {
public static function main() {
var target = new Some();
var binding = new Bindable(target);
binding.one = 5;
// binding.two = 0.12; // intentionally unset field
binding.three = []; // wrong type
binding.four = 'str'; // no such field in wrapped class
trace(binding.one, binding.two, binding.three, binding.four, binding.five);
// outputs: 5, null, [], str, null
trace(target.one, target.two, target.three);
// outputs: 5, null, []
}
}
class Some {
public var one:Int;
public var two:Float;
public var three:Bool;
public function new() {}
}
abstract Bindable<TClass>(TClass) {
public inline function new(source) { this = source; }
#:op(a.b) public function setField<T>(name:String, value:T) {
Reflect.setField(this, name, value);
// TODO notify handlers
return value;
}
#:op(a.b) public function getField<T>(name:String):T {
return cast Reflect.field(this, name);
}
}
So I have some frustrating issues: interface of wrapped object doesn't expose to wrapper, so there's no auto completion or strict type checking, some necessary attributes can be easily omitted or even misspelled.
Is it possible to fix my solution or should I better move to the macros?
I almost suggested here to open an issue regarding this problem. Because some time ago, there was a #:followWithAbstracts meta available for abstracts, which could be (or maybe was?) used to forward fields and call #:op(a.b) at the same time. But that's not really necessary, Haxe is powerful enough already.
abstract Binding<TClass>(TClass) {
public function new(source:TClass) { this = source; }
#:op(a.b) public function setField<T>(name:String, value:T) {
Reflect.setField(this, name, value);
// TODO notify handlers
trace("set: $name -> $value");
return value;
}
#:op(a.b) public function getField<T>(name:String):T {
trace("get: $name");
return cast Reflect.field(this, name);
}
}
#:forward
#:multiType
abstract Bindable<TClass>(TClass) {
public function new(source:TClass);
#:to function to(t:TClass) return new Binding(t);
}
We use here multiType abstract to forward fields, but resolved type is actually regular abstract. In effect, you have completion working and #:op(a.b) called at the same time.
You need #:forward meta on your abstract. However, this will not make auto-completion working unless you remove #:op(A.B) because it shadows forwarded fields.
EDIT: it seems that shadowing happened first time I added #:forward to your abstract, afterwards auto-completion worked just fine.

TypeScript - passing a class as an argument, and reflection

I am writing a generic unmarshaller. It converts graph DB data to generated TypeScript (1.8.7) model classes. The input is JSON. The output should be an instance of a model class.
My ultimate goal is to create something like Hibernate OGM, only for Tinkerpop Frames and TypeScript, with REST endpoint in the middle.
What's the right way to pass a class as a parameter and reach it's static members? I want to have something like this:
SomeModel some = <SomeModel> unmarshaller.fromJSON({/*Object from JSON*/}, SomeModel);
I've tried to write a method.
Not sure if I am heading in the right direction, feel free to suggest different approaches.
public fromJSON(input: Object, clazz: typeof FrameModel): FrameModel
{
// This only demonstrates access to Framemodel's metadata
// about original Java model classes.
clazz.graphPropertyMapping;
clazz.graphRelationMapping;
let result = {};
...
return result;
}
...
But when I tried to execute this on Plunker, I got execution errors with unuseful stacktrace.
The model superclass looks like this:
/**
* Things common to all Frames models on the Typescript side.
*/
export class FrameModel
{
// Model metadata
static discriminator: string;
static graphPropertyMapping: { [key:string]:string; };
static graphRelationMapping: { [key:string]:string; };
// Each instance needs a vertex ID
private vertexId: number;
public getVertexId(): number {
return this.vertexId;
}
}
Sample model class:
import {TestPlanetModel} from './TestPlanetModel';
import {TestShipModel} from './TestShipModel';
export class TestGeneratorModel extends FrameModel
{
static discriminator: string = 'TestGenerator';
static graphPropertyMapping: { [key:string]:string; } = {
bar: 'boo',
name: 'name',
rank: 'rank',
};
static graphRelationMapping: { [key:string]:string; } = {
colonizes: 'colonizedPlanet',
commands: 'ship',
};
boo: string;
name: string;
rank: string;
public colonizedPlanet: TestPlanetModel[]; // edge label 'colonizedPlanet'
public ship: TestShipModel; // edge label 'ship'
}
I haven't found much material on reflection and class handling in TypeScript.
I know how I would do this in Java.
I know how I would do this in JavaScript.
I understand that I might achieve similar results with decorators, but having fields or static fields seemed a bit simpler, for generated models.
You've maybe already noticed that class members cannot have const keyword. But you could go with static instead. Also member should be public if you want it to be accessible from outside world.
public static graphPropertyMapping: { [key:string]:string; } = {
bar: 'boo',
name: 'name',
rank: 'rank',
};
As for creating result instance:
let result = new clazz();
//copy properties
return result;
If I understand you correctly then here's something to help you get started:
interface Model {}
interface ModelData {}
interface MyModelConstructor<M extends Model, D extends ModelData> {
new(data: D): M;
// static members
graphPropertyMapping: any;
graphRelationMapping: any;
}
class Unmarshaller {
public fromJSON<T>(input: string | ModelData, ctor: MyModelConstructor<T, ModelData>): T {
let data: ModelData = (typeof input === "string") ? JSON.parse(input) : input;
let propertyMapping = ctor.graphPropertyMapping;
let relationMapping = ctor.graphRelationMapping;
// do whatever with the mappings
return new ctor(input);
}
}
(code in playground)
I don't know how your models look like, so I hope this is close enough.
I recently released an enhanced version of the TypeScript compiler that allows exactly what you are expecting: read all (static or not) fields metadata from a class. For example you can write:
interface MyInterface {
active:boolean;
description: string;
}
class MyClass {
id: number;
name: string;
myComplexField: MyInterface;
}
function printMembers(clazz: Class) {
let fields = clazz.members.filter(m => m.type.kind !== 'function'); //exclude methods.
for(let field of fields) {
let typeName = field.type.kind;
if(typeName === 'class' || typeName === 'interface') {
typeName = (<Class | Interface>field.type).name;
}
console.log(`Field ${field.name} of ${clazz.name} has type: ${typeName}`);
}
}
printMembers(MyClass.getClass());
this is the output:
$ node main.js
Field id of MyClass has type: number
Field name of MyClass has type: string
Field myComplexField of MyClass has type: MyInterface
Of course, if you change the members property access of clazz to statics you will retrieve all static members. These information can be accessed at coding time too, so you can use autocompletion.
You can do the same with Interfaces metadata. Simply write MyInterface for example, and access its members.
You can find the project here.

Adding observer for KVO without pointers using Swift

In Objective-C, I would normally use something like this:
static NSString *kViewTransformChanged = #"view transform changed";
// or
static const void *kViewTransformChanged = &kViewTransformChanged;
[clearContentView addObserver:self
forKeyPath:#"transform"
options:NSKeyValueObservingOptionNew
context:&kViewTransformChanged];
I have two overloaded methods to choose from to add an observer for KVO with the only difference being the context argument:
clearContentView.addObserver(observer: NSObject?, forKeyPath: String?, options: NSKeyValueObservingOptions, context: CMutableVoidPointer)
clearContentView.addObserver(observer: NSObject?, forKeyPath: String?, options: NSKeyValueObservingOptions, kvoContext: KVOContext)
With Swift not using pointers, I'm not sure how to dereference a pointer to use the first method.
If I create my own KVOContext constant for use with the second method, I wind up with it asking for this:
let test:KVOContext = KVOContext.fromVoidContext(context: CMutableVoidPointer)
EDIT: What is the difference between CMutableVoidPointer and KVOContext? Can someone give me an example how how to use them both and when I would use one over the other?
EDIT #2: A dev at Apple just posted this to the forums: KVOContext is going away; using a global reference as your context is the way to go right now.
There is now a technique officially recommended in the documentation, which is to create a private mutable variable and use its address as the context.
(Updated for Swift 3 on 2017-01-09)
// Set up non-zero-sized storage. We don't intend to mutate this variable,
// but it needs to be `var` so we can pass its address in as UnsafeMutablePointer.
private static var myContext = 0
// NOTE: `static` is not necessary if you want it to be a global variable
observee.addObserver(self, forKeyPath: …, options: [], context: &MyClass.myContext)
override func observeValue(forKeyPath keyPath: String?, of object: Any?, change: [NSKeyValueChangeKey: Any]?, context: UnsafeMutableRawPointer?) {
if context == &myContext {
…
}
else {
super.observeValue(forKeyPath: keyPath, of: object, change: change, context: context)
}
}
Now that KVOContext is gone in Xcode 6 beta 3, you can do the following. Define a global (i.e. not a class property) like so:
let myContext = UnsafePointer<()>()
Add an observer:
observee.addObserver(observer, forKeyPath: …, options: nil, context: myContext)
In the observer:
override func observeValueForKeyPath(keyPath: String!, ofObject object: AnyObject!, change: [NSObject : AnyObject]!, context: UnsafePointer<()>) {
if context == myContext {
…
} else {
super.observeValueForKeyPath(keyPath, ofObject: object, change: change, context: context)
}
}
Swift 4 - observing contentSize change on UITableViewController popover to fix incorrect size
I had been searching for an answer to change to a block based KVO because I was getting a swiftlint warning and it took me piecing quite a few different answers together to get to the right solution. Swiftlint warning:
Block Based KVO Violation: Prefer the new block based KVO API with keypaths when using Swift 3.2 or later. (block_based_kvo).
My use case was to present a popover controller attached to a button in a Nav bar in a view controller and then resize the popover once it's showing - otherwise it would be too big and not fitting the contents of the popover. The popover itself was a UITableViewController that contained static cells, and it was displayed via a Storyboard segue with style popover.
To setup the block based observer, you need the following code inside your popover UITableViewController:
// class level variable to store the statusObserver
private var statusObserver: NSKeyValueObservation?
// Create the observer inside viewWillAppear
override func viewWillAppear(_ animated: Bool) {
super.viewWillAppear(animated)
statusObserver = tableView.observe(\UITableView.contentSize,
changeHandler: { [ weak self ] (theTableView, _) in self?.popoverPresentationController?.presentedViewController.preferredContentSize = theTableView.contentSize
})
}
// Don't forget to remove the observer when the popover is dismissed.
override func viewDidDisappear(_ animated: Bool) {
if let observer = statusObserver {
observer.invalidate()
statusObserver = nil
}
super.viewDidDisappear(animated)
}
I didn't need the previous value when the observer was triggered, so left out the options: [.new, .old] when creating the observer.
Update for Swift 4
Context is not required for block-based observer function and existing #keyPath() syntax is replaced with smart keypath to achieve swift type safety.
class EventOvserverDemo {
var statusObserver:NSKeyValueObservation?
var objectToObserve:UIView?
func registerAddObserver() -> Void {
statusObserver = objectToObserve?.observe(\UIView.tag, options: [.new, .old], changeHandler: {[weak self] (player, change) in
if let tag = change.newValue {
// observed changed value and do the task here on change.
}
})
}
func unregisterObserver() -> Void {
if let sObserver = statusObserver {
sObserver.invalidate()
statusObserver = nil
}
}
}
Complete example using Swift:
//
// AppDelegate.swift
// Photos-MediaFramework-swift
//
// Created by Phurg on 11/11/16.
//
// Displays URLs for all photos in Photos Library
//
// #see http://stackoverflow.com/questions/30144547/programmatic-access-to-the-photos-library-on-mac-os-x-photokit-photos-framewo
//
import Cocoa
import MediaLibrary
// For KVO: https://developer.apple.com/library/content/documentation/Swift/Conceptual/BuildingCocoaApps/AdoptingCocoaDesignPatterns.html#//apple_ref/doc/uid/TP40014216-CH7-ID12
private var mediaLibraryLoaded = 1
private var rootMediaGroupLoaded = 2
private var mediaObjectsLoaded = 3
#NSApplicationMain
class AppDelegate: NSObject, NSApplicationDelegate {
#IBOutlet weak var window: NSWindow!
var mediaLibrary : MLMediaLibrary!
var allPhotosAlbum : MLMediaGroup!
func applicationDidFinishLaunching(_ aNotification: Notification) {
NSLog("applicationDidFinishLaunching:");
let options:[String:Any] = [
MLMediaLoadSourceTypesKey: MLMediaSourceType.image.rawValue, // Can't be Swift enum
MLMediaLoadIncludeSourcesKey: [MLMediaSourcePhotosIdentifier], // Array
]
self.mediaLibrary = MLMediaLibrary(options:options)
NSLog("applicationDidFinishLaunching: mediaLibrary=%#", self.mediaLibrary);
self.mediaLibrary.addObserver(self, forKeyPath:"mediaSources", options:[], context:&mediaLibraryLoaded)
NSLog("applicationDidFinishLaunching: added mediaSources observer");
// Force load
self.mediaLibrary.mediaSources?[MLMediaSourcePhotosIdentifier]
NSLog("applicationDidFinishLaunching: done");
}
override func observeValue(forKeyPath keyPath: String?, of object: Any?, change: [NSKeyValueChangeKey : Any]?, context: UnsafeMutableRawPointer?) {
NSLog("observeValue: keyPath=%#", keyPath!)
let mediaSource:MLMediaSource = self.mediaLibrary.mediaSources![MLMediaSourcePhotosIdentifier]!
if (context == &mediaLibraryLoaded) {
NSLog("observeValue: mediaLibraryLoaded")
mediaSource.addObserver(self, forKeyPath:"rootMediaGroup", options:[], context:&rootMediaGroupLoaded)
// Force load
mediaSource.rootMediaGroup
} else if (context == &rootMediaGroupLoaded) {
NSLog("observeValue: rootMediaGroupLoaded")
let albums:MLMediaGroup = mediaSource.mediaGroup(forIdentifier:"TopLevelAlbums")!
for album in albums.childGroups! {
let albumIdentifier:String = album.attributes["identifier"] as! String
if (albumIdentifier == "allPhotosAlbum") {
self.allPhotosAlbum = album
album.addObserver(self, forKeyPath:"mediaObjects", options:[], context:&mediaObjectsLoaded)
// Force load
album.mediaObjects
}
}
} else if (context == &mediaObjectsLoaded) {
NSLog("observeValue: mediaObjectsLoaded")
let mediaObjects:[MLMediaObject] = self.allPhotosAlbum.mediaObjects!
for mediaObject in mediaObjects {
let url:URL? = mediaObject.url
// URL does not extend NSObject, so can't be passed to NSLog; use string interpolation
NSLog("%#", "\(url)")
}
}
}
}

Understanding OOP in Actionscript

A.as :
public class A {
public function getFunction():Function {
return function():void {
if(this is C) {
trace("C");
} else {
trace("not C");
}
}
}
public function func1():void {
var internalFunc:Function = getFunction();
internalFunc();
}
}
B.as :
public class B extends A implements C {
}
In some other class :
var b:B = new B();
B.func1();
Output is :
"Not C"
I was expecting the trace output to be
"C"
Can someone explain why?
An anonymous function, if called directly, is scoped to the global object. If you trace this inside it, you will see [object global] instead of [object B], as you would, if this refered to b.
A common workaround is using a closure:
var self:A = this;
return function():void {
if(self is C) {
trace("C");
} else {
trace("not C");
}
}
Please note however, the instance-members of a class defining an anonymous function are available from within. This works, because they are resolved at compile time.
edit in response to Amarghosh's question:
Yes, this points to the global object, but that doesn't mean, you cannot access the instance members of the declaring class. This little piece of code should explain the details:
package {
import flash.display.Sprite;
public class Test extends Sprite {
private var foo:String = "foo";
public function Test() {
var anonymous:Function = function ():void {
trace(foo);//foo
trace(this.foo);//undefined
};
anonymous();
}
}
}
greetz
back2dos
A few things with the code that I assume are just typos?
The getFunction() method doesn't return anything and will thus cause a compiler error.
Your call code calls func1() as a static method, not as a method on an instance of the B. This will also cause a compiler error. I believe these are typos.
In my tests, using your modified code. The output is C. There must be something else going on with your code. Here are my mods to A:
public function getFunction():Function {
if(this is C) {
trace("C");
} else {
trace("not C");
}
return getFunction;
}
Here is my mod to the runnable code, which I put in creationComplete of an empty MXML Application file:
var b:B = new B();
b.func1();
I assume your "real world" code is more extensive than the sample and there must be something else going on.

Resources