As getAdvancedTypefaceMetrics of SkTypefce is now private, In my old code it is using in below format.
int FontPlatformData::emSizeInFontUnits() const
{
if (m_emSizeInFontUnits)
return m_emSizeInFontUnits;
SkAdvancedTypefaceMetrics* metrics = 0;
if (m_typeface)
metrics = m_typeface->getAdvancedTypefaceMetrics(SkAdvancedTypefaceMetrics::kNo_PerGlyphInfo);
if (metrics) {
m_emSizeInFontUnits = metrics->fEmSize;
metrics->unref();
} else
m_emSizeInFontUnits = 1000; // default value copied from Skia.
return m_emSizeInFontUnits;
}
is there any alternative provided for this API in SKia.
SkTypeFace.h has "getUnitsPerEm()" method, that should help.
Related
I've tried to search for instance caching and singletons on Google and StackOverflow without success, seeing only posts about module.exports, if you know a post that answers this question, feel free to reference it. Thank you!
I have an application that needs to work on a set of objects that rarely change, and hence need to be cached for performance optimisation.
Here is a toy example where a single property is set directly.
When I call the application, I export an object that will contain the set of cached objects in assets_cached.js:
const Assets = {};
module.exports.Assets = Assets;
In another module of the application I have an ES6 class:
const _ = require('lodash')
const { Assets } = require('./assets_cached')
class Asset {
constructor(id, some_property) {
if (id in Assets) {
// Update instance data with cached properties
_.assign(this, Assets_cached[id]);
} else {
// If it's not cached, create a new object
this.id = id;
this.some_property = some_property;
// Cache this object
Assets_cached[id] = this;
}
}
getProperty() {
return this.some_property;
}
setProperty(value) {
this.some_property = value;
// Is there a way of avoiding having to do this double assignment?
Assets_cached[id].some_property = value;
}
}
module.exports = Asset;
How may I avoid having to set the some_property twice (in the current instance and the cache, while ensuring that other instances are updated in parallel)?
Ideally I'd like to do something like:
if (id in Assets) {
this = Assets.cached[id]
}
inside the constructor, but this is not possible.
What's the most elegant and correct way of making this work?
Ideally I'd like to do something like this = Assets.cached[id] inside the constructor
The magic keyword here is return. You can just return an arbitrary object from the constructor and it will be used instead of this.
constructor(id, some_property) {
if (id in Assets) {
// use cached instance instead of creating a new one
return Assets_cached[id];
} else {
this.id = id;
this.some_property = some_property;
// Cache this object
Assets_cached[id] = this;
}
}
Here is the approach to the comment that was made some half an hour ago ...
const { Assets_cached } = require('./assets_cached');
// const { AssetStore } = require('./assetstore');
class Asset {
constructor(id, some_property) { // clean/lean constructor.
this.id = id;
this.some_property = some_property;
}
getProperty() {
return this.some_property;
}
setProperty(value) {
this.some_property = value;
}
}
function isAsset(type) {
// poor man's approach ... change to something more feasible.
return (type instanceof Asset);
}
function createAsset(id, some_property) { // factory that also handles caching.
var
asset = Assets_cached[id];
// asset = AssetStore.get(id);
if (!(asset && isAsset(asset))) {
asset = Assets_cached[id] = (new Asset(id, some_property));
// AssetStore.put(id, (asset = new Asset(id, some_property)));
}
return asset;
}
module.exports = {
create : createAsset,
isAsset : isAsset
};
Note
One also should consider providing a minimal API to Assets_cached, something like put/set, get and delete instead of Assets_cached being an entirely exposed, plain key-value store.
Here is my code
public static void save(IgniteContext igniteContext, String cacheName, Dataset<Row> dataSet) {
CacheConfiguration<BinaryObject, BinaryObject> cacheConfiguration = new CacheConfiguration<BinaryObject, BinaryObject>(cacheName)
.setAtomicityMode(CacheAtomicityMode.ATOMIC)
.setBackups(0)
.setAffinity(new RendezvousAffinityFunction(false, 2))
.setIndexedTypes(BinaryObject.class, BinaryObject.class);
IgniteCache<BinaryObject, BinaryObject> rddCache = igniteContext.ignite()
.getOrCreateCache(cacheConfiguration)
.withKeepBinary();
rddCache.clear();
IgniteRDD<BinaryObject, BinaryObject> igniteRDD = igniteContext.fromCache(cacheName);
StructField[] fields = dataSet.schema().fields();
RDD<BinaryObject> binaryObjectJavaRDD = dataSet.toJavaRDD().map(row -> {
BinaryObjectBuilder valueBuilder = igniteContext.ignite().binary().builder(BinaryObject.class.getCanonicalName());
for (int i = 0; i < fields.length; i++) {
valueBuilder.setField(fields[i].name(), convertValue(String.valueOf(row.get(i)), fields[i].dataType())); //convertValue converts value to specific datatype
}
return valueBuilder.build();
}).rdd();
igniteRDD.saveValues(binaryObjectJavaRDD);
}
I have a problem with the above code, that is even after successful completion of this method cache remains empty. Dataset has 20 rows so that is not the problem.
The other problem is that if I use savePairs method from IgniteRDD then I have to generate the Key by myself(here Key is BinaryObject), so how to do that?
update
saveDFInPairs(IgniteContext igniteContext, Dataset<Row> dataSet, IgniteRDD<BinaryObject, BinaryObject> igniteRDD) {
StructField[] fields = dataSet.schema().fields();
JavaRDD<Tuple2<BinaryObject, BinaryObject>> rdd = dataSet.toJavaRDD().map(row -> {
BinaryObjectBuilder keyBuilder = igniteContext.ignite()
.binary().builder("TypeName");
keyBuilder.setField("id", row.mkString().hashCode());
BinaryObject key = keyBuilder.build();
BinaryObjectBuilder valueBuilder = igniteContext.ignite()
.binary().builder("TypeName");
for (int i = 0; i < fields.length; i++) {
valueBuilder.setField(fields[i].name(), convert(row, i, fields[i].dataType()));
}
BinaryObject value = valueBuilder.build();
return new Tuple2<>(key, value);
});
igniteRDD.savePairs(rdd.rdd(), true);
}
Couple of considerations:
The type name (the one passed to the builder() method) should be a meaningful name representing the data type. Do not use BinaryObject class name for this.
setIndexedTypes(BinaryObject.class, BinaryObject.class) is incorrect. This should specify classes to be processed for query annotations. If you don't have classes, you can use QueryEntity to configure queries. See this page for further details: https://apacheignite.readme.io/docs/sql-queries
Other than that code looks correct. I would recommend to try with default settings and check if it works this way. Also it's not very clear how you check that the data is in cache or not.
This is a bit foreign to me and I'm probably not understanding it correctly. This is what I have:
var imgModule = (function() {
var imgLocations = {};
var images = [];
imgLocations.setImage = function(img, location) {
imgLocations[img] = location;
}
imgLocations.getImg = function(img) {
return imgLocations[img];
}
imgLocations.setImageArray = function(img) {
images.push(img);
}
imgLocations.getImageArray = function() {
return images;
}
return imgLocations;
}());
I want to be able to access the imgLocations Object and images array from outside this function. The setting functions work, but
document.getElementById("but").onclick = function() {
console.log(imgModule.imgLocations.getImageArray());
console.log(imgModule.imgLocations.getImg(imgName));
}
Both return "undefined". How do I access these variables? And how can I improve this function? Please be patient with me and explain what I'm doing wrong :) I'm trying to learn it the right way instead of defining a global variable outside all functions.
The reason why this isn't working, is because your imgModule is returning the imgLocations object. That being the case, imgModule will actually be the imgLocations object. So you would access your methods like so:
imgModule.setImage()
imgModule.getImg()
imgModule.getImageArray()
imgModule.setImageArray()
And as #gillesc stated. If you are wanting to keep the current syntax of imgModule.imgLocations.getImg() then you could return the imgLocations like so
return {
imgLocations: imgLocations
}
doing so would allow you to add more functionality to your module
return {
imgLocations: imgLocations,
otherObject: otherObject
}
...
imgModule.otherObject.someFunctionCall();
The problem is you are returning the object created and are not setting it as a property of an object.
So in your case this is how it would work.
document.getElementById("but").onclick = function() {
console.log(imgModule.getImageArray());
console.log(imgModule.getImg(imgName));
}
What you need to do is return it like this
return {
imgLocations: imgLocations
}
If you want the API you are attending to create and still have access to the array which you can not do currently.
You don't access imgModule.imgLocations, since what you return is imgLocations, you should access them as:
document.getElementById("but").onclick = function() {
console.log(imgModule.getImageArray());
console.log(imgModule.getImg(imgName));
}
It seems you try to write module pattern.
For deep understanding, I recommend you following article:
The Module Pattern, by Addy Osmani
and pay attention to example with counter:
var testModule = (function () {
var counter = 0;
return {
incrementCounter: function () {
return counter++;
},
resetCounter: function () {
console.log( "counter value prior to reset: " + counter );
counter = 0;
}
};
})();
// Usage:
// Increment our counter
testModule.incrementCounter();
// Check the counter value and reset
// Outputs: counter value prior to reset: 1
testModule.resetCounter();
Update
Since the problem has been found I've also find out that Box2D for web is leaking on every side :/
To show this I made a simple circle moving in a static polygon and here is the result after some time.
Notice how the following items are leaking as I'm not creating any body or changing the world in any way:
b2Vec2
Features
b2ManifoldPoint
b2ContactID
b2Manifold
b2ContactEdge
b2PolyAndCircleContact
Array
...
Original post
I have a problem because I'm profiling my game and the garbage collector doesnt' delete my bodies, contacts and other stuff. Then I've looked at what are they keeping from the GC and was the Box2D itself. This might lead to 2 options: I'm doing it bad or Box2D is leaking. I consider is my cause.
What exactly is keeping it?
contact.m_nodeA.other was appearing to be the most used to keep it from GC.
other times: m_fixtureB in a contact... see image
You can see that the body has a __destroyed property. That is set manually before deleting it with world.DestroyBody(body)
When I destroy a body I call it after I call the step method on the world.
As you can see from the box2d method it doesn't get rid of the other variable nor it changes it to another body and my body is not GC.
Any idea of what I'm missing here?
Now I can fix the problem only if the world.Step is not ran:
var gravity = new Box2D.Vec2(0, 0);
var doSleep = true;
var world = new Box2D.World(gravity, doSleep);
var step = false;
var fixtureDef = new Box2D.FixtureDef();
fixtureDef.density = 1.0;
fixtureDef.friction = 0.5;
fixtureDef.restitution = 0.2;
fixtureDef.shape = new Box2D.PolygonShape();
fixtureDef.shape.SetAsBox(1, 1);
var bodyDef = new Box2D.BodyDef;
bodyDef.type = Box2D.Body.b2_dynamicBody;
bodyDef.position.x = 0.4;
bodyDef.position.y = 0.4;
var bodies = []
var fix = [];
window.c = function(){
for(var i = 0; i < 100; i++){
var body = world.CreateBody(bodyDef);
body._id = i;
fix.push(body.CreateFixture(fixtureDef));
bodies.push(body);
}
if(step){world.Step(1/60, 3, 3); world.ClearForces();}
console.log('Created', bodies)
fixtureDef = null;
bodyDef = null;
}
window.d = function(){
_.each(bodies, function(body, i){
body.DestroyFixture(fix[i]);
world.DestroyBody(body);
fix[i] = null;
bodies[i] = null;
})
if(step){world.Step(1/60, 3, 3); world.ClearForces();}
bodies = null;
fix = null;
}
Change the step to true and the memory leak problem appears again.
Reproduce the memory leak problem:
Code in your file:
var gravity = new Box2D.Vec2(0, 0);
var doSleep = true;
var world = new Box2D.World(gravity, doSleep);
var bodies = []
window.c = function(){
for(var i = 0; i < 100; i++){
var bodyDef = new Box2D.BodyDef();
bodyDef.type = 2;
var shape = new Box2D.PolygonShape();
shape.SetAsBox(1, 1);
var fixtureDef = new Box2D.FixtureDef();
fixtureDef.shape = shape;
var body = world.CreateBody(bodyDef);
body._id = i;
body.CreateFixture(fixtureDef);
bodies.push(body);
}
world.Step(0.3, 3, 3);
console.log('Created', bodies)
}
window.d = function(){
_.each(bodies, function(body, i){
world.DestroyBody(body);
bodies[i] = null;
})
world.Step(0.3, 3, 3);
bodies = null;
}
Open google chrome:
Then open your profile and make a snapshot.
Now run the c() method in your console to create 100 bodies
Now snapshot 2
Search in snapshot for b2Body and you'll find 100 Object count
Now run d() to delete all your bodies;
Force Garbage collection by clicking on the garbage can
Make a snapshot 3
Search for b2Body and you'll also find 100 Object count
At the last step should only be 0 objects as they have been destroyed. Instead of this you'll find this:
Now you can see there are a lot of references from b2ContactEdge. Now if you remove the world.Step part of the code you will only see 2 references to the body.
If you remove this line
body.CreateFixture(fixtureDef);
or making the body static is not leaking anymore.
My game loop
...gameLoop = function(o){
// used a lot here
var world = o.world;
// calculate the new positions
var worldStepSeconds = o.worldStepMs / 1000;
// step world
world.Step(worldStepSeconds, o.velocityIterations, o.positionIterations)
// render debug
if(o.renderDebug){
world.DrawDebugData();
}
// always to not accumulate forces, maybe some bug occurs
world.ClearForces();
// tick all ticking entities
_.each(o.getTickEntitiesFn(), function(actor){
if(!actor) return;
actor.tick(o.worldStepMs, o.lastFrameMs);
})
// update PIXI entities
var body = world.GetBodyList();
var worldScale = world.SCALE;
var destroyBody = world.DestroyBody.bind(world);
while(body){
var actor = null;
var visualEntity = null;
var box2DEntity = o.getBox2DEntityByIdFn(body.GetUserData());
if(box2DEntity){
visualEntity = o.getVisualEntityByIdFn(box2DEntity.getVisualEntityId());
if(box2DEntity.isDestroying()){
// optimization
body.__destroyed = true;
world.DestroyBody(body);
box2DEntity.completeDestroy();
}
}
if(visualEntity){
if(visualEntity.isDestroying()){
visualEntity.completeDestroy();
}else{
var inverseY = true;
var bodyDetails = Utils.getScreenPositionAndRotationOfBody(world, body, inverseY);
visualEntity.updateSprite(bodyDetails.x, bodyDetails.y, bodyDetails.rotation);
}
}
// this delegates out functionality for each body processed
if(o.triggersFn.eachBody) o.triggersFn.eachBody(world, body, visualEntity);
body = body.GetNext();
}
// when a joint is created is then also created it's visual counterpart and then set to userData.
var joint = world.GetJointList();
while(joint){
var pixiGraphics = joint.GetUserData();
if(pixiGraphics){
// In order to draw a distance joint we need to know the start and end positions.
// The joint saves the global (yes) anchor positions for each body.
// After that we need to scale to our screen and invert y axis.
var anchorA = joint.GetAnchorA();
var anchorB = joint.GetAnchorB();
var screenPositionA = anchorA.Copy();
var screenPositionB = anchorB.Copy();
// scale
screenPositionA.Multiply(world.SCALE);
screenPositionB.Multiply(world.SCALE);
// invert y
screenPositionA.y = world.CANVAS_HEIGHT - screenPositionA.y
screenPositionB.y = world.CANVAS_HEIGHT - screenPositionB.y
// draw a black line
pixiGraphics.clear();
pixiGraphics.lineStyle(1, 0x000000, 0.7);
pixiGraphics.moveTo(screenPositionA.x, screenPositionA.y);
pixiGraphics.lineTo(screenPositionB.x, screenPositionB.y);
}
joint = joint.GetNext();
}
// render the PIXI scene
if(o.renderPixi){
o.renderer.render(o.stage)
}
// render next frame
requestAnimFrame(o.requestAnimFrameFn);
}
Code from Box2d:
b2ContactManager.prototype.Destroy = function (c) {
var fixtureA = c.GetFixtureA();
var fixtureB = c.GetFixtureB();
var bodyA = fixtureA.GetBody();
var bodyB = fixtureB.GetBody();
if (c.IsTouching()) {
this.m_contactListener.EndContact(c);
}
if (c.m_prev) {
c.m_prev.m_next = c.m_next;
}
if (c.m_next) {
c.m_next.m_prev = c.m_prev;
}
if (c == this.m_world.m_contactList) {
this.m_world.m_contactList = c.m_next;
}
if (c.m_nodeA.prev) {
c.m_nodeA.prev.next = c.m_nodeA.next;
}
if (c.m_nodeA.next) {
c.m_nodeA.next.prev = c.m_nodeA.prev;
}
if (c.m_nodeA == bodyA.m_contactList) {
bodyA.m_contactList = c.m_nodeA.next;
}
if (c.m_nodeB.prev) {
c.m_nodeB.prev.next = c.m_nodeB.next;
}
if (c.m_nodeB.next) {
c.m_nodeB.next.prev = c.m_nodeB.prev;
}
if (c.m_nodeB == bodyB.m_contactList) {
bodyB.m_contactList = c.m_nodeB.next;
}
this.m_contactFactory.Destroy(c);
--this.m_contactCount;
}
b2ContactFactory.prototype.Destroy = function (contact) {
if (contact.m_manifold.m_pointCount > 0) {
contact.m_fixtureA.m_body.SetAwake(true);
contact.m_fixtureB.m_body.SetAwake(true);
}
var type1 = parseInt(contact.m_fixtureA.GetType());
var type2 = parseInt(contact.m_fixtureB.GetType());
var reg = this.m_registers[type1][type2];
if (true) {
reg.poolCount++;
contact.m_next = reg.pool;
reg.pool = contact;
}
var destroyFcn = reg.destroyFcn;
destroyFcn(contact, this.m_allocator);
}
I have the same problem, but I think I find out from where it comes.
Instead of m_* try functions, like GetFixtureA() instead of m_fixtureA.
Totti did you ever figure this out? It looks like box2dweb requires manual destruction and memory management.
I think I have found your leaks, un-implemented ( static class ) destruction functions:
b2Joint.Destroy = function (joint, allocator) {}
b2CircleContact.Destroy = function (contact, allocator) {}<
b2PolygonContact.Destroy = function (contact, allocator) {}
b2EdgeAndCircleContact.Destroy = function (contact, allocator) {}<
b2PolyAndCircleContact.Destroy = function (contact, allocator) {}
b2PolyAndEdgeContact.Destroy = function (contact, allocator) {}
[UPDATE...]
b2DestructionListener.b2DestructionListener = function () {};
b2DestructionListener.prototype.SayGoodbyeJoint = function (joint) {}
b2DestructionListener.prototype.SayGoodbyeFixture = function (fixture) {}
b2Contact.prototype.Reset(fixtureA, fixtureB)
called with with one/both fixture arguments resets passed in fixture/s BUT ALSO pass in NO arguments and it 'nulls' all the the b2Contact properties! (UNTESTED:) but I suggest set your YOURcontactListener class up to handle all contact callbacks EVERY call with Reset(??) dynamically configureable as logic requies EVERY call (there are more than you'd imagine each and every world step).
Also take Colt McAnlis clever advice and strategically pre allocate all the memory the life of your game will need (by creating game and box2d object pools now you know objects can be reset) so the garbage collector NEVER runs until, you destroy object pools at times of your own convenience.... i.e when you close the tab, or your device needs recharging! ;D [...UPDATE]
// you can define and assign your own contact listener ...via...
YOUR.b2world.b2ContactManager.m_world.m_contactList = new YOURcontactlistener();<br>[edit]...if you dont it actually does have Box2D.Dynamics.b2ContactListener.b2_defaultListener.
// box2d in the worldStep calls YOURcontactlistener.update() via:
this.b2world.b2ContactManager.m_world.m_contactList.Update(this.m_contactListener) // this.m_contactListener being YOURS || b2_defaultListener;
// which instantiates ALL your listed leaking object like so:
{b2Contact which instantiates {b2ContactEdge} and {b2Manifold which instantiates {b2ManifoldPoint{which instantiates m_id.key == ContactID{which instantiates Features}}}} along with {B2Vec2} are instantiated in b2ContactResult ...which I can not actually find but assume it must be instantiated in the Solver.
// There is a Contacts.destroyFcn callback is CREATED in....
b2ContactFactory.prototype.Destroy = function (contact) {...}
// then Contacts.destroyFcn callback(s) are privately REGISTERED in....
b2ContactFactory.prototype.InitializeRegisters() {...}
...via...
this.AddType = function (createFcn, destroyFcn, type1, type2) {...}
...BUT... THOSE privately registered ARE four of the un-implimented static class function from above...
b2PolygonContact.Destroy = function (contact, allocator) {}
b2EdgeAndCircleContact.Destroy = function (contact, allocator) {}
b2PolyAndCircleContact.Destroy = function (contact, allocator) {}
b2PolyAndEdgeContact.Destroy = function (contact, allocator) {}
So I havn't tested it yet but it looks like box2dweb just gives you the Destroy callback/handler functions and you have to read the source to find all the properties you need to null. [Edit] In combination with b2Contact.prototype.Reset(fixtureA, fixtureB)
But either way pretty confident the functions above(possibly incomplete) are callback/handlers, and can be used to null your way back to performance for anyone else who stumbles across this problem. Pretty sure Totti's moved on(dont forget to handle your 'this' scope in callbacks).
Hi can some one please tell me how to copy one data store to another in dojo. I tried it in following way but it doesn't work. Here I'm try to copy data from jsonStore to newGridStore.
jsonStore.fetch({query:{} , onComplete: onComplete});
var onComplete = function (items, request) {
newGridStore = null;
newGridStore = new dojo.data.ItemFileWriteStore({
data : {}
});
if (items && items.length > 0) {
var i;
for (i = 0; i < items.length; i++) {
var item = items[i];
var attributes = jsonStore.getAttributes(item);
if (attributes && attributes.length > 0) {
var j;
for (j = 0; j < attributes.length; j++) {
var newItem = {};
var values = jsonStore.getValues(item, attributes[j]);
if (values) {
if (values.length > 1) {
// Create a copy.
newItem[attributes[j]] = values.slice(0, values.length);
} else {
newItem[attributes[j]] = values[0];
}
}
}
newGridStore.newItem(newItem);
}
}
}
}
Based on the comments asked above. You are trying to copy values to a new Store for the single reason to be able to detect which values have changes and then save them individually, without having to send the entire store.
This approach is totally wrong.
Dojo has isDirty() and offers you the ability to revert() a store back to it's original values. It knows which values have changed and you don't need to do this.
Take a look at the bog standard IFWS here: http://docs.dojocampus.org/dojo/data/ItemFileWriteStore
Make sure you read everything from here: http://docs.dojocampus.org/dojo/data/ItemFileWriteStore#id8
What you want to do is create your own _saveCustom method which you will override your store with, and then when you save, you will be able to see which values have changed.
Click on the demo at the very bottom of the page. It shows you EXACTLY how do to it using _saveCustom