Well, the problem itself is kind of hard to describe briefly, so here's a live example to demonstrate. It seems like I'm misunderstanding something about how Rx.js works, otherwise the functionality here comes from a bug.
What I tried to do was a simple reactive rendering setup, where what you see on the screen, and what events happen are both described in terms of Observables. The problem is that, for some indiscernible reason, the events are dropped entirely when the code is written one way, yet work fine with code that should theoretically be equivalent.
So, let's start with the first case in the example code above:
var dom = makeBox('one');
var clicks = Rx.Observable.fromEvent(dom, 'click');
If you create a DOM fragment, then you can simply use fromEvent to get an Observable for whatever event it emits. So far, so good. You can click this box and see a bunch of lines written to the log.
Now, the next step would be to make the DOM reactive, to express how it changes over time.
var domStream = Rx.Observable.return(makeBox('two'));
var clicks = domStream.flatMapLatest(function(dom) {
return Rx.Observable.fromEvent(dom, 'click');
});
That would make it an Observable, using return here to produce the simplest, constant case. The events you're interested in would be the ones emitted by the latest version of the dom, and that's exactly what the flatMapLatest operator does. This variant still works.
Ultimately, the goal would be to generate the current DOM state based on some application state. That is, map it from one Observable to another. Let's go with the simplest version for now, have a single constant value as the state, and then map it to the same fixed output we used previously:
var updates = Rx.Observable.return(1);
var domStream = updates.map(function (update) {
return makeBox('three');
});
var clicks = domStream.flatMapLatest(function(dom) {
return Rx.Observable.fromEvent(dom, 'click');
});
This should not be any different from the previous version. However, this outputs no events, no matter what you do.
What exactly is going on here? Did I misunderstand some fundamental concept of Rx, or what? I've run into some issues with hot vs cold Observables, but that seems unrelated in this minimal case. So, I'm kind of out of ideas. Can anyone enlighten me?
Sorry to tell you but it is a Hot vs Cold issue.
It is a subtle issue, but the difference between
Rx.Observable.return(makeBox('two'))
and
Rx.Observable.return(1).map(function() {return makeBox('three'); })
Is that the first returns a constant every time you subscribe to it, that is,
a box that you created initially. The second returns a new box every time the Observable is subscribed to, this causes a problem since you actually subscribe to the domStream variable twice, you are creating two instances of Box three, one which has event handlers but isn't shown and one that does not and is shown.
The fix is that you either need to use approach 2 or you need to convert the third into a hot stream either by using:
domStream.replay(1).refCount()
Or by using
domStream.publish()
then after all subscriptions are completed:
domStream.connect()
Related
I am trying to process an insert event from the CKEditor 5.
editor.document.on("change", (eventInfo, type, data) => {
switch (type) {
case "insert":
console.log(type, data);
break;
}
});
When typing in the editor the call back is called. The data argument in the event callback looks like approximately like this:
{
range: {
start: {
root: { ... },
path: [0, 14]
},
end: {
root: { ... },
path: [0, 15]
}
}
}
I don't see a convenient way to figure out what text was actually inserted. I can call data.range.root.getNodeByPath(data.range.start.path); which seems to get me the text node that the text was inserted in. Should we then look at the text node's data field? Should we assume that the last item in the path is always an offset for the start and end of the range and use that to substring? I think the insert event is also fired for inserting non-text type things (e.g. element). How would we know that this is indeed a text type of an event?
Is there something I am missing, or is there just a different way to do this all together?
First, let me describe how you would do it currently (Jan 2018). Please, keep in mind that CKEditor 5 is now undergoing a big refactoring and things will change. At the end, I will describe how it will look like after we finish this refactoring. You may skip to the later part if you don't mind waiting some more time for the refactoring to come to an end.
EDIT: The 1.0.0-beta.1 was released on 15th of March, so you can jump to the "Since March 2018" section.
Until March 2018 (up to 1.0.0-alpha.2)
(If you need to learn more about some class API or an event, please check out the docs.)
Your best bet would be simply to iterate through the inserted range.
let data = '';
for ( const child of data.range.getItems() ) {
if ( child.is( 'textProxy' ) ) {
data += child.data;
}
}
Note, that a TextProxy instance is always returned when you iterate through the range, even if the whole Text node is included in the range.
(You can read more about stringifying a range in CKEditor5 & Angular2 - Getting exact position of caret on click inside editor to grab data.)
Keep in mind, that InsertOperation may insert multiple nodes of a different kind. Mostly, these are just singular characters or elements, but more nodes can be provided. That's why there is no additional data.item or similar property in data. There could be data.items but those would just be same as Array.from( data.range.getItems() ).
Doing changes on Document#change
You haven't mentioned what you want to do with this information afterwards. Getting the range's content is easy, but if you'd like to somehow react to these changes and change the model, then you need to be careful. When the change event is fired, there might be already more changes enqueued. For example:
more changes can come at once from collaboration service,
a different feature might have already reacted to the same change and enqueued its changes which might make the model different.
If you know exactly what set of features you will use, you may just stick with what I proposed. Just remember that any change you do on the model should be done in a Document#enqueueChanges() block (otherwise, it won't be rendered).
If you would like to have this solution bulletproof, you probably would have to do this:
While iterating over data.range children, if you found a TextProxy, create a LiveRange spanning over that node.
Then, in a enqueueChanges() block, iterate through stored LiveRanges and through their children.
Do your logic for each found TextProxy instance.
Remember to destroy() all the LiveRanges afterwards.
As you can see this seems unnecessarily complicated. There are some drawbacks of providing an open and flexible framework, like CKE5, and having in mind all the edge cases is one of them. However it is true, that it could be simpler, that's why we started refactoring in the first place.
Since March 2018 (starting from 1.0.0-beta.1)
The big change coming in 1.0.0-beta.1 will be the introduction of the model.Differ class, revamped events structure and a new API for big part of the model.
First of all, Document#event:change will be fired after all enqueueChange blocks have finished. This means that you won't have to be worried whether another change won't mess up with the change that you are reacting to in your callback.
Also, engine.Document#registerPostFixer() method will be added and you will be able to use it to register callbacks. change event still will be available, but there will be slight differences between change event and registerPostFixer (we will cover them in a guide and docs).
Second, you will have access to a model.Differ instance, which will store a diff between the model state before the first change and the model state at the moment when you want to react to the changes. You will iterate through all diff items and check what exactly and where has changed.
Other than that, a lot of other changes will be conducted in the refactoring and below code snippet will also reflect them. So, in the new world, it will look like this:
editor.document.registerPostFixer( writer => {
const changes = editor.document.differ.getChanges();
for ( const entry of changes ) {
if ( entry.type == 'insert' && entry.name == '$text' ) {
// Use `writer` to do your logic here.
// `entry` also contains `length` and `position` properties.
}
}
} );
In terms of code, it might be a bit more of it than in the first snippet, but:
The first snippet was incomplete.
There are a lot fewer edge cases to think about in the new approach.
The new approach is easier to grasp - you have all the changes available after they are all done, instead of reacting to a change when other changes are queued and may mess up with the model.
The writer is an object that will be used to do changes on the model (instead of Document#batch API). It will have methods like insertText(), insertElement(), remove(), etc.
You can check model.Differ API and tests already as they are already available on master branch. (The internal code will change, but API will stay as it is.)
#Szymon Cofalik's answer went into a direction "How to apply some changes based on a change listener". This made it far more complex than what's needed to get the text from the Document#change event, which boils down to the following snippet:
let data = '';
for ( const child of data.range.getChildren() ) {
if ( child.is( 'textProxy' ) ) {
data += child.data;
}
}
However, reacting to a change is a tricky task and, therefore, make sure to read Szymon's insightful answer if you plan to do so.
So, I recently discovered EaselJS (and more generally CreateJS) and I'm trying to figure out a way to make turbomedia (ie this kind of thing) with it.
At the current time, I'm working on reversibility. A turbomedia tells its story through a series of states/frames, and a key feature is the ability to move back and forth between these frames at will (usually through keystrokes). In order to achieve this property of reversibility, I need states to be independent from previous events (ie state #2 must be the same whether it's reached from state #1 or state #3).
Until recently, I'd simply work with single bitmaps (such that each state would correspond to one existing file) so the problem would never present itself. However, now I'd like the ability to have states be compositions made out of multiple images (since this allows a lot more flexibility). Thus, a state might be described by the array ["sky3", "ground2", "character5"], meaning "this state contains the images stored in sky3, ground2 and character5".
The problem I'm hitting is twofold.
First, I need the ability to compare array contents so that whenever the current state changes, the new state is compared with the previous one and images are swapped around as needed (ie going from ["sky1", "kid1"] to ["sky2", "kid1"] will remove sky1 from the stage, add sky2, and keep kid1 since it's present in both states). This is to preserve animation timings across states, and to try and make transitions lighter (although I'm not sure that's needed?).
But I have no idea how to compare arrays contents like this.
The second problem is probably much simpler, but I lack experience with Javascript and honestly I have no idea what I'm doing wrong. I am unable to target the content of my states. Here is my init():
stage = new createjs.Stage("testing");
currentstate = 1;
terra1 = new createjs.Bitmap("terra1.png");
terra2 = new createjs.Bitmap("terra2.png");
bullet1 = new createjs.Bitmap("bullet1.png");
bullet2 = new createjs.Bitmap("bullet2.png");
state1 = ["terra1"];
state2 = ["terra2", "bullet1"];
state3 = ["terra2", "bullet2"];
calcstate = "state" + currentstate;
// Call the first state (at least that's what I'm going for).
console.log(calcstate);
// This returns "state1". I want it to return ["terra1"] since that's the
//content of state1.
for (i = 0; i < calcstate.length; i++) {
stage.addChild(calcstate[i]);
// Currently useless since previous code doesn't work, but would be the
// function to "create the first stage".
};
stage.update();
So yeah, for now I'm pretty much stuck. Any suggestion?
You are not referring to the instances properly.
Your calcState will be a string (such as "state1"), and not a reference to the variable state1. You could use bracket access to reference it:
Example:
this[calcState]
// OR, depending on your scope
window[calcState]
Even if your reference the state arrays correctly, they just contain strings themselves, so you would be adding "terra1" to the stage, and not the instance terra1. You can use bracket access here too, but a better way is to actually add the instances to your state arrays instead:
Example:
state1 = [terra1];
state2 = [terra2, bullet1];
state3 = [terra2, bullet2];
I recommend using console.log() to output the values calcState, as well as the calcstate[i] in your for loop, which should shed some light at what your are looking at.
An easier way to handle this would to make a states array, which has sub-elements:
states = [
[terra1],
[terra2, bullet1],
[terra2, bullet2]
];
// Refer to your states. Note that calcState should be 0-2 and not 1-3
states[calcState][i];
Hope that helps.
I am trying to use ngGrid to make somewhat of a "tree-control" which I can build dynamically by calling API's. ngGrid allows for grouping on rows, yet the nature of it requires that all rows be present at the beginning. This is unfortunate for the fact that an API to pull back all generation data for a File Integrity Monitoring system would be insanely slow and stupid. Instead, I wish to build the "tree" dynamically on the expansion of each generation.
I am trying to inject children (ngRows) into a group-row (ngAggregate) on a callback, yet I do not think that I am calling the correct constructor for the ngRows for the fact that the rows are ignored by the control
Through the use of the aggregateTemplate option on the gridOptions for ngGrid, I have been able to intersept the expansion of a group quite easily.
(maybe not easily, but still)
I've replaced the ng-click of the default template:
ng-click="row.toggleExpand()"
with:
ng-click="$parent.$parent.rowExpanded(row)"
I know that it's a bit of a hack, but we can get to that later. For now, it gets the job done.
The way that I discovered how to work my way up the $scope to my rowExpanded function was by setting a breakpoint in ngGrid's "row.toggleExpand" function and calling it from the template as so:
ng-click="row.toggleExpand(this)"
Once I retrieve the group I want, I call an API to get the children for said group. I then need to make the return as children of the row. I decided to do this by calling ngGrid's ngRow factory:
row.children = [];
for(var i = 0; i < childData.length; i++)
{
row.children[row.children.length] = row.rowFactory.buildEntityRow(childData[i], i);
}
row.toggleExpand();
... yet this does not appear to be working. The rows are not showing up after I do the expand! Why won't my rows show up?
Here's my current Plunker!
By the way
I've placed a debugger statement within the group-expand callback. As long as you have your debugger open, you should catch a breakpoint on the expansion of a group.
Thanks everybody!
I found my answer, I'm an idiot....
I got this control working, and then realized that it was a total hack, that I could have used the control the way it was meant to be used and it would have worked much better, had much better work-flow, and it would have saved me an entire day of development. If you are wondering how you use the control this way, the answer is that you don't.
I got the stupid thing to work by updating my data structure after the round trip and forcing the grid to refresh, pretty obvious. I had to set the grid options so that groups were always expanded and I had to control the collapser icon logic myself, outside of ngGrid. I never called row.toggleExpand. I also hid any rows with null values by a function call within an ng-if on my rowTemplate. After all that was said and done, I put my foot in my mouth.
I was doing this test case to see how much using the this selector speeds up a process. While doing it, I decided to try out pre-saved element variables as well, assuming they would be even faster. Using an element variable saved before the test appears to be the slowest, quite to my confusion. I though only having to "find" the element once would immensely speed up the process. Why is this not the case?
Here are my tests from fastest to slowest, in case anyone can't load it:
1
$("#bar").click(function(){
$(this).width($(this).width()+100);
});
$("#bar").trigger( "click" );
2
$("#bar").click(function(){
$("#bar").width($("#bar").width()+100);
});
$("#bar").trigger( "click" );
3
var bar = $("#bar");
bar.click(function(){
bar.width(bar.width()+100);
});
bar.trigger( "click" );
4
par.click(function(){
par.width(par.width()+100);
});
par.trigger( "click" );
I'd have assumed the order would go 4, 3, 1, 2 in order of which one has to use the selector to "find" the variable more often.
UPDATE: I have a theory, though I'd like someone to verify this if possible. I'm guessing that on click, it has to reference the variable, instead of just the element, which slows it down.
Fixed test case: http://jsperf.com/this-vs-thatjames/10
TL;DR: Number of click handlers executed in each test grows because the element is not reset between tests.
The biggest problem with testing for micro-optimizations is that you have to be very very careful with what you're testing. There are many cases where the testing code interferes with what you're testing. Here is an example from Vyacheslav Egorov of a test that "proves" multiplication is almost instantaneous in JavaScript because the testing loop is removed entirely by the JavaScript compiler:
// I am using Benchmark.js API as if I would run it in the d8.
Benchmark.prototype.setup = function() {
function multiply(x,y) {
return x*y;
}
};
var suite = new Benchmark.Suite;
suite.add('multiply', function() {
var a = Math.round(Math.random()*100),
b = Math.round(Math.random()*100);
for(var i = 0; i < 10000; i++) {
multiply(a,b);
}
})
Since you're already aware there is something counter-intuitive going on, you should pay extra care.
First of all, you're not testing selectors there. Your testing code is doing: zero or more selectors, depending on the test, a function creation (which in some cases is a closure, others it is not), assignment as the click handler and triggering of the jQuery event system.
Also, the element you're testing on is changing between tests. It's obvious that the width in one test is more than the width in the test before. That isn't the biggest problem though. The problem is that the element in one test has X click handlers associated. The element in the next test has X+1 click handlers.
So when you trigger the click handlers for the last test, you also trigger the click handlers associated in all the tests before, making it much slower than tests made earlier.
I fixed the jsPerf, but keep in mind that it still doesn't test just the selector performance. Still, the most important factor that skewes the results is eliminated.
Note: There are some slides and a video about doing good performance testing with jsPerf, focused on common pitfalls that you should avoid. Main ideas:
don't define functions in the tests, do it in the setup/preparation phase
keep the test code as simple as possible
compare things that do the same thing or be upfront about it
test what you intend to test, not the setup code
isolate the tests, reset the state after/before each test
no randomness. mock it if you need it
be aware of browser optimizations (dead code removal, etc)
You don't really test the performance between the different techniques.
If you look at the output of the console for this modified test:
http://jsperf.com/this-vs-thatjames/8
You will see how many event listeners are attached to the #bar object.
And you will see that they are not removed at the beginning for each test.
So the following tests will always become slower as the previous ones because the trigger function has to call all the previous callbacks.
Some of this increase in slowness is because the object reference is already found in memory, so the compiler doesn't have to go looking in memory for the variable
$("#bar").click(function(){
$(this).width($(this).width()+100); // Only has to check the function call
}); // each time, not search the whole memory
as opposed to
var bar = $("#bar");
...
bar.click(function(){
bar.width(bar.width()+100); // Has to search the memory to find it
}); // each time it is used
As zerkms said, dereferencing (having to look up the memory reference as I describe above) has some but little effect on the performance
Thus the main source of slowness in difference for the tests you have performed is the fact that the DOM is not reset between each function call. In actuality, a saved selector performs just about as fast as this
Looks like the performance results you're getting has nothing to do with the code. If you look at these edited tests, you can see that having the same code in two of the tests (first and last) yield totally different results.
I don't know, but if I had to guess I would say it is due to concurrency and multithreading.
When you do $(...) you call the jQuery constructor and create a new object that gets stored in the memory. However, when you reference to an existing variable you do not create a new object (duh).
Although I have no source to quote I believe that every javascript event gets called in its own thread so events don't interfere with eachother. By this logic the compiler would have to get a lock on the variable in order to use it, which might take time.
Once again, I am not sure. Very interesting test btw!
I am trying to bind a property of an object to a property that's bound in an ArrayController. I want all of this to occur after the object has already been created and added to the ArrayController.
Here is a fiddle with a simplified example of what I'm trying to achieve.
I am wondering if I'm having problems with scope - I've already tried to bind to the global path (i.e. 'App.objectTwoController.objectOne.param3') to set the binding to. I've also tried to bind directly to the objectOneController (which is not what I want to do, but tried it just to see if it worked) and that still didn't work.
Any ideas on what I'm doing incorrectly? Thanks in advance for taking the time to look at this post.
So in the example below (I simplified it a little bit, but same principles apply)... The method below ends up looking for "objectOne" on "objectTwo" instead of on the "objectTwoController".
var objectTwoController: Em.Object.create({
objectOneBinding: 'App.objectOne',
objectTwoBinding: 'App.objectTwo',
_onSomething: function() {
var objectTwo = this.get('objectTwo');
objectTwo.bind('param2', Em.Binding.from('objectOne.param3'));
}.observes('something')
});
The problem is that you can't bind between two none relative objects. If you look in the "connect" method in ember you will see that it only takes one reference object (this) in which to observe both paths (this is true for 9.8.1 from your example and the ember-pre-1.0 release).
You have few options (that I can think of at least).
First: You can tell the objects about each other and in turn the relative paths will start working. This will actually give "objectTwo" an object to reference when binding paths.
....
objectTwo.set('objectOne', this.get('objectOne');
....
Second: You could add your own observer/computed property that will just keep the two in sync (but it is a little more verbose). You might be able to pull off something really slick but it maybe difficult. Even go so far as writing your own binding (like Transforms) to allow you to bind two non-related objects as long as you have paths to both.
_param3: function(){
this.setPath('objectTwo.param2', this.getPath('objectOne.param3');
}.observes('objectOne.param3')
You can make these dynamically and not need to pre-define them...
Third: Simply make them global paths; "App.objectOneController.content.param3" should work as your binding "_from" path (but not sure how much this helps you in your real application, because with larger applications I personally don't like everything global).
EDIT: When setting the full paths. Make sure you wait until end of the current cycle before fetching the value because bindings don't always update until everything is flushed. Meaning, your alert message needs to be wrapped in Ember.run.next or you will not see the change.