I use Reflux.connect methods to change component state, but I can't get the different part in shouldComponentUpdate between nextState and this.state. Actually, this.state already has the new values I was expecting to be in nextState when shouldComponentUpdate is called.
What should I need to do when I want to use shouldComponentUpdate with Reflux?
var component = React.createClass({
mixins: [Reflux.connect(store1, 'store1data'),
Reflux.connect(store2, 'store2data')],
shouldComponentUpdate: function(nextProps, nextState) {
//Here this.state.store1data is equal to nextState.store1data
}
Make sure you're not mutating the state and instead returning a new copy of it. If your store only changes the field inside the state, then the this.state pointer already points to the mutated state.
so in your store1, instead of:
store1Data.someKey = newValue;
this.trigger(store1Data)
try doing:
store1Data = _.assign({}, store1Data, {someKey: newValue}); // this uses lodash, for example
this.trigger(store1Data)
That way you will get a fresh copy of store1Data every time it changes.
Immutability is an important concept when working with states, especially in React/Flux, since it allows you to determine quickly whether a state was changed. Try making a habit of always returning a new copy when changing the state, and to never change to anything inside the state unless you clone it first.
Immutablity in ReactImmutableJs
Related
After clicking the "Return" action on a selected item from a completed Purchase Receipt, we're trying to get the "Open PO Line" value to default to false. Anyone know how customize this?
The field defaulting seems to be overwritten when we press the "Return" button. We've tried several different events in the grid but none of the seem to work.
The desired result is to default the "Open PO Line" to false after a return and once the return is released the Purchase Order line associated with the return should remain completed.
Research
The AllowOpen field on POReceiptLine is a PXBool. This means that the value must be populated via a PXDBScalar, PXFormula, etc. or via some business logic in the graph. To see what is happening, let's look at the DAC for POReceiptLine...
#region AllowOpen
public abstract class allowOpen : PX.Data.BQL.BqlBool.Field<allowOpen> { }
protected Boolean? _AllowOpen;
[PXBool()]
[PXUIField(DisplayName = "Open PO Line", Visibility = PXUIVisibility.Service, Visible = true)]
public virtual Boolean? AllowOpen
{
get
{
return this._AllowOpen;
}
set
{
this._AllowOpen = value;
}
}
#endregion
As you can see, there isn't any logic to this field in the DAC, so we need to turn to the graph to see how it is used. (Even if there were logic in the DAC, we would need to see if the graph does something. However, logic in the DAC might have been an easy override with CacheAttached - unfortunately, not in this case.)
Let's turn to POReceiptEntry where the return is handled. We find AllowComplete and AllowOpen being set in the POReceiptLine_RowSelected event, as we would expect since it must be populated on the graph side of code having no logic in the DAC.
if ((row.AllowComplete == null || row.AllowOpen == null) && fromPO)
{
POLineR source = PXSelect<POLineR,
Where<POLineR.orderType, Equal<Required<POLineR.orderType>>,
And<POLineR.orderNbr, Equal<Required<POLineR.orderNbr>>,
And<POLineR.lineNbr, Equal<Required<POLineR.lineNbr>>>>>>
.Select(this, row.POType, row.PONbr, row.POLineNbr);
// Acuminator disable once PX1047 RowChangesInEventHandlersForbiddenForArgs [Legacy, moved the exception here from PX.Objects.acuminator because the condition was changed]
row.AllowComplete = row.AllowOpen = (row.Released == true) ?
(row.ReceiptType != POReceiptType.POReturn ? source?.Completed == true : source?.Completed != true) :
(source?.AllowComplete ?? false);
The field is populated in the row.AllowComplete = row.AllowOpen = (row.Released == true) ?... section of code.
Subsequently, we see that the CopyFromOrigReceiptLine method sets this value to false on the "destLine" being created.
destLine.AllowOpen = false;
As that isn't "true" then we know this isn't our spot. Continuing on, we see in UpdatePOLineCompletedFlag that AllowComplete and AllowOpen are being set. This could be our spot (or one of them).
row.AllowComplete = row.AllowOpen = poLineCurrent.AllowComplete;
Side note: It is worth noting that this line appears twice in an if then else. In both cases it is going to be executed, therefore it would be better coding practice to place this identical statement AFTER the if then else since both the if and else conditions will execute this same statement regardless of the if.
This particular use appears to be pulling the value from the AllowComplete field of the PO Line being received. At this point, you should consider if you need to look upstream at the PO Line to see if the field there needs to be manipulated as well. I cannot answer that for you as your business case will drive the decision.
There also is a line in the Copy method that sets AllowComplete and AllowOpen.
aDest.AllowComplete = aDest.AllowOpen = (aSrc.CompletePOLine == CompletePOLineTypes.Quantity);
The Copy method is overloaded, and the other signature of the method sets the values to true.
aDest.AllowComplete = true;
aDest.AllowOpen = true;
Both of these cases may need customization as well, but I don't think it's the primary issue.
Next Steps
At this point, we see that either the field is being set in UpdatePOLineCompletedFlag or in methods that seem related to copying records. You will need to investigate further if the copy related methods warrant a change as well. However, I think the initial focus should be on the UpdatePOLineCompletedFlag method.
If we find the other points identified require customization, we likely will handle them all the same way... Override the base method/event, invoke the original method/event in our override, and then force the values to fit our business case. Careful testing will be needed since forcibly altering these values may create unforeseen negative ripples.
Something to try
We want to update (or create) a graph extension for POReceiptEntry to override the UpdatePOLineCompleteFlag method. This compiles, but it is completely untested on my part. We need to create a delegate and specify the PXOverride attribute. Then we want to execute the base method before we override the field(s) in question.
Note the extra code (commented out) as a reminder that you need to be careful of methods (typically events) updating our record in the cache and needing to be located so that we don't use a stale copy of the record. I don't think that's necessary in this case, but it seems to be somewhat obscure in code samples that I see. Of course, that is because I'm always looking at the code repository which rarely has graph extensions overriding event handlers!
#region CreateReceiptEntry Override
public delegate void UpdatePOLineCompleteFlagDelegate(POReceiptLine row, bool isDeleted, POLine aOriginLine);
[PXOverride]
public virtual void UpdatePOLineCompleteFlag(POReceiptLine row, bool isDeleted, POLine aOriginLine, UpdatePOLineCompleteFlagDelegate baseMethod)
{
//Execute original logic
baseMethod(row, isDeleted, aOriginLine);
/* If the base method has updated the cache, then we would need to locate the updated record in the cache to proceed
* This tends to be the case more often with event handlers, so it probably isn't needed in this case.
* This is just for reference as a training reminder
//If row has been updatd in the baseMethod, let's go get the updated cache values
POReceiptLine locateRow = Base.transactions.Locate(row);
if (locateRow != null) row = locateRow;
*/
//Override the fields to false - need to test to see if this creates any issues with breaking existing business logic
row.AllowComplete = row.AllowOpen = false;
}
#endregion
If this doesn't get you the specific answer you need, I hope it at least gives you some insight into how to hunt down "the spot" to change. I suspect you may need to update the POLine for a complete solution as hinted above. (See the event handler POReceiptLine_AllowOpen_FieldUpdated for the code that leads me to that conclusion.)
Good luck with your customization, and happy coding!
So imagine the following scenario, using the Parse platform on iOS:
I get a PFObject from the server, let's call it GlassChalice.
Someone else, let's say Bill Blofeld, changes GlassChalice from a different location.
Later, I make some changes to my local GlassChalice, but don't save them to the server.
Still later, I want to update GlassChalice, but I want to update it to the current server values, in other word Bill Blofeld's values. I do not want to replace the server values with my local values, and also do not want to reset my local values to the values GlassChalice was loaded with.
So if I use revert(), will I get what I want?
According to the Parse docs:
- revert Clears any changes to this object made since the last call to save and sets it back to the server state.
...but, as in my example, clearing "changes made since the last call to save" and setting it "back to the server state" aren't always the same thing.
So far this seems like the only way to guarantee the results I want, but it has one obvious problem:
public func updateObjectFromServer(_ objectToUpdate: PFObject, then doThis: (()->Void)? = nil) {
let query = PFObject.query()
query?.whereKey("objectId", equalTo: objectToUpdate.objectId!)
query?.getFirstObjectInBackground (block: {
(serverObject, error) in
if error.isNil() {
objectToUpdate["numberOfLimbs"] = serverObject?["numberOfLimbs"]
objectToUpdate["eyePlacement"] = serverObject?["eyePlacement"]
objectToUpdate["crossStitchingTalentRating"] = serverObject?["crossStitchingTalentRating"]
objectToUpdate["clamsEaten"] = serverObject?["clamsEaten"]
} else {
//handle error...
}
doThis?()
})
}
But the huge problem here is that I have to know all the key names, and type them in explicitly, for this to work.
Is there a better, more generic, way?
How can I determine, at runtime, if a list in Dart is a "fixed list" ?
There are (at least) three ways to create a fixed-length list in Dart:
var fixed = new List(5); // fixed at five elements
var alsoFixed = new List.filled(5, null); // fixed at five elements, set to null
var fixedToo = new List.from([1,2,3], growable: false);
How do I ask, in code, if fixed, alsoFixed, and fixedToo are fixed-length?
You can try to add an element and remove it to know if the list has a fixed length:
bool hasFixLength(List list) {
try {
list
..add(null)
..removeLast();
return false;
} on UnsupportedError {
return true;
}
}
runtimeType can be used for that, but shouldn't because it's not very reliable especially with dart2js otherwise there isn't. AFAIR there were several requests to add support for this but was declined.
We did not add any way to see if a list is fixed (or const).
So far, we have only encountered one use case where this would fit into well-written code: checking that a function returned a mutable array. We decided that this use case was not worth changing the API.
Most of the time, an additional boolean flag that asks to modify an existing list is cleaner (since then, the caller can decide what the callee should do).
If you have another use case, please let us know and we will investigate.
I've never used actionscript before, and but I've just had to dive into it in order to get a map working.
I'm using the following code to add a map marker, replacing a previous one if one exists:
public var tracer:Array = new Array();
public var tracerLng:Number = 0;
for ( var i : Number=1 ; i<64000 ; i++)
{
//Check if there is already a marker, if so get rid of it
if(tracerLng > 0) {
map.removeOverlay(tracer[0]);
tracer[0] = null;
tracer.pop();
}
// Set up a marker
var trackMrk:Marker = new Marker(
new LatLng(_lat, _lng),
new MarkerOptions({
strokeStyle: new StrokeStyle({color: 0x987654}),
fillStyle: new FillStyle({color: 0x223344, alpha: 0.8}),
radius: 12,
hasShadow: true
})
);
//Add the marker to the array and show it on the map
tracerLng = tracer.push(trackMrk);
map.addOverlay(tracer[0]);
}
My first problem is that running this code (The 64000 repeats are for testing, the final application won't need to be run quite THAT many times). Either way, memory usage increases by about 4kB/s - how do I avoid that happening?
Secondly - could anyone advise me on how to make that program more graceful?
Thanks in advance for advice
This isn't a memory leak, it's probably the result of created events - enter frames, mouse events, custom events etc. Provided that your memory doesn't keep going up and up forever, it's nothing to be worried about - it'll get garbage collected in due course.
Some points on your code:
The tracer Array doesn't seem to do anything - you only seem to be holding one thing in there at a time, so an array makes no sense. If you need an Array, use Vector instead. It's smaller and faster. More so if you create one with a specific length.
Don't create a new Marker unless you need one. Reuse old objects. Learn about object pooling: http://help.adobe.com/en_US/as3/mobile/WS948100b6829bd5a6-19cd3c2412513c24bce-8000.html or http://lostinactionscript.com/2008/10/30/object-pooling-in-as3/
The LatLng and MarkerOptions (including the stroke and fill objects) don't seem to change (I'm assuming the LatLng object lets you set a new position). If that's the case, don't create new ones when you don't need to. If you need to create new ones, StrokeStyle and FillStyle seem good candidates for a "create once, use everywhere" policy.
Create a destroy() function or similar in your Marker class and explicitly call it when you need to delete one (just before setting it to null or popping it from the array). In the destroy() function, null out any parameters to non-base classes (int, Number, String etc). Garbage collection runs using a reference counting method and a mark and sweep method. Ideally, you want to run everything using reference counting as it's collected quicker and stops any stalls in your program.
I explain memory management in AS3 a bit more here: http://divillysausages.com/blog/tracking_memory_leaks_in_as3
Also included is a class that helps you track down memory leaks if there are any
In my responder I have:
Spanish.ADDWORD = SC.Responder.create({
didBecomeFirstResponder: function(responder) {
var store = this._store = Spanish.store.chain(); //buffer changes
var word = store.createRecord(Spanish.Word, {word: "", english: ""});
Spanish.addWordController.set("content",word);
//show the dialog
var pane = Spanish.getPath('addWordPage.mainPane');
pane.append();
pane.makeFirstResponder();
},
submit: function(){
this._store.commitChanges().destroy();
Spanish.makeFirstResponder(Spanish.READY);
}
}
Before I had the DataSource hooked-up, and I was using local, everything worked. When I click submit now no new object is created and createRecord is not being called.
A possible problem is that you are calling .destroy() immediately. This shouldn't be an issue, but as you said it was working while using fixtures (which are synchronous). Now that you are using a dataSource (which is usually asynchronous), it may be getting interrupted. Try removing the .destory(), and see if that resolves your issue.
Another option to try, is that there may be a bug in the nested store, in that if you create a new record (rather than edit an existing one), the 'did it change' test may fail (as there is nothing to compair it too), so calling commitChanges(YES) will force a commit, without the check.