Develop Reference

Unity Raycast results different on two devices

I use the following codes to detect which UI element is the player pointing. It works well in Unity Editor, on one of my Android phone, but failed on a second Android phone.
public static T RaycastOnFirstPointed<T>(EventSystem eventSystem, GraphicRaycaster raycaster)
{
if (Application.platform == RuntimePlatform.Android || Application.platform == RuntimePlatform.IPhonePlayer)
{
if (Input.touches == null || Input.touches.Length == 0) return default;
}
var m_PointerEventData = new PointerEventData(eventSystem);
m_PointerEventData.position = Input.mousePosition;
var results = new List<RaycastResult>();
raycaster.Raycast(m_PointerEventData, results); // On the 2nd phone the results.length is 0
foreach (RaycastResult result in results)
{
var component = result.gameObject.GetComponent<T>();
if (component != null)
{
return component;
}
}
return default;
}
This method is called within a IEndDragHandler.OnEndDrag method. I tried to debug it on the 2nd Android phone, and saw the results.length is 0 after Raycast was executed.(it should be > 0 since I ended my drag on several overlapped UI elements, and it is > 0 on the 1st phone, not sure why the 2nd phone is different)
I don't have any clue where should I go from here to find the problem. Please show me some directions, thank you!
More info:
Phone 1: MI 8 Lite, OS: MIUI 10.3
Phone 2: MI 9, OS: MIUI 10.2.35
============Update===========
Reason found:
When IEndDragHandler.OnEndDrag is triggered, the Input.mousePosition has different value on the two devices. 1st Phone has the touch position from the last frame, while the 2nd phone seems clears that value, and it has a very large wrong value. I'm working on how to solve this in a proper way. Any suggestions are welcomed.

Problem solved. As I updated in the question, this problem is caused by wrong Input.mousePosition value on different devices. So I updated the RaycastOnFirstPointed method as follows:
public static T RaycastOnFirstPointed<T>(EventSystem eventSystem, GraphicRaycaster raycaster, PointerEventData eventData = null)
{
if (Application.platform == RuntimePlatform.Android || Application.platform == RuntimePlatform.IPhonePlayer)
{
if (Input.touches == null || Input.touches.Length == 0) return default;
}
var m_PointerEventData = new PointerEventData(eventSystem);
if (eventData == null)
{
m_PointerEventData.position = Input.mousePosition;
}
else
{
m_PointerEventData.position = eventData.position;
}
var results = new List<RaycastResult>();
raycaster.Raycast(m_PointerEventData, results);
foreach (RaycastResult result in results)
{
var component = result.gameObject.GetComponent<T>();
if (component != null)
{
return component;
}
}
return default;
}
When call this method within OnEndDrag, pass in the PointerEventData that unity provides and use its position.

Chrome 84 Webview update causes stuttering while rendering MediaStream - Samsung Tab A 8.0

This was all working just fine until the Chrome 84 update, with 85 not fixing it either. Disabling chrome altogether on the impacted devices is my only current workaround. I am trying to render the camera preview from a Samsung Galaxy Tab A inside a cordova app.
First, I fetch the devices and select the rear facing camera if there are multiple:
navigator.mediaDevices.enumerateDevices().then(function(dev) {
ctrl.devices = dev.filter(function(el) { return el.kind == "videoinput"; });
if (ctrl.devices.length > 1) {
// default to the back facing camera
ctrl.selectedDevice = ctrl.devices[1];
ctrl.startCamera();
}
else if (ctrl.devices.length == 1) {
ctrl.selectedDevice = ctrl.devices[0];
ctrl.startCamera();
}
else {
ctrl.NoDeviceFound = true;
console.log("No camera found!");
}
});
Once I have a device selected, I start the camera up with this function:
ctrl.startCamera = function startCamera() {
ctrl.stopCamera(); //this function just stops the stream if one was already open
if (ctrl.selectedDevice) {
var constraints = {video: { deviceId: { exact: ctrl.selectedDevice.deviceId } }};
navigator.mediaDevices.getUserMedia(constraints).then(handleSuccess).catch(handleError);
}
}
My success handler is where the stream then gets injected into the video element and rendered:
function handleSuccess(stream) {
ctrl.stream = stream;
window.stream = stream; // make stream available to browser console
window.video = angular.element("video")[0];
if (window.screen.orientation.type.indexOf('landscape') == 0) {
video.width = angular.element("div.modal-body").width();
video.height = video.width / 1.33;
}
else {
video.height = angular.element("div.modal-body").height();
video.width = video.height * .75
}
if (typeof video.srcObject != "undefined") {
video.srcObject = stream;
}
else {
video.src = window.URL.createObjectURL(stream);
}
video.play();
}
I have a simple <video autoplay></video> element as my target.
Here's a video of what the preview looks like
It ends up just being the preview that is broken though. You can see that it's stuck on the first frame there, however if I go ahead and capture an image it will accurately reflect where the camera is pointed despite the rendered stream being broken. Occasionally instead of getting stuck on that first frame, it will sort of 'boomerang' between the first few frames.
Edit: I believe this issue is being tracked here

It was in fact this chromium bug, which has been fixed for Chrome 86 and confirmed working on my devices.

Digital persona SDK - native problems

I am using in my app Digital Persona SDK for fingerprint identification.
When i use the identify function on less then 250 fmds it works fine.
Engine.Candidate candidates[] = m_engine.Identify(searchedFmd, 0, fmdArray, DEFAULT_THRESHOLD, 1); //fmdArray < 250
But with fmdArray > 250 it gives me a native runtime error:
A/art: art/runtime/indirect_reference_table.cc:132] JNI ERROR (app bug): local reference table overflow (max=512)
Now i runned this app on couple of android devices and came to conclusion that my app crushes with fmdArray > 250 when its running on android 7. But android 8 works fine. In 8 i can preform a check on even 4000 fmds and it works fine.
But i need to run this code in a specific device, that running android 7.
I tried to run it in couple of threads of 250 fmds only. But after single run there is another problem with the SDK. On the second run it doesnt works.
This is what i do:
First i get a fingerprint capture that i want to identify:
Reader.CaptureResult capture = m_reader.Capture(fidFormat, UrUSDK.DefaultImageProcessing, m_DPI, timeout);
// In second run, code after this line is not executed.
// My guees its not coming back from native. No exeptions. No errors.
...
Fmd scannedFmd = m_engine.CreateFmd(capture.image, fmdFormat);
...
int index = identifyFinger(fmds, scannedFmd);
...
private int identifyFinger(List<Fmd> fmdSearchArray, Fmd scannedFmd) {
List<List<Fmd>> lists = splitToChunks(fmdSearchArray);
AtomicInteger index = new AtomicInteger(-1);
List<Callable<Void>> threads = new ArrayList<>(lists.size());
AtomicInteger iteratorIndex = new AtomicInteger(0);
for (int i = 0; i < lists.size(); i++) {
int currentChunk = i;
Callable<Void> thread = () -> {
System.out.println(Thread.currentThread().getName() + " with chunk: " + iteratorIndex.getAndIncrement());
Fmd[] fmds = lists.get(currentChunk).toArray(new Fmd[IDENTIFY_BOUNDARY]);
try {
Engine.Candidate[] candidates = m_engine.Identify(scannedFmd, 0, fmds, threshold, 1);
if (candidates.length > 0) {
index.set(candidates[0].fmd_index + (currentChunk * IDENTIFY_BOUNDARY));
}
} catch (UareUException e) {
}
System.out.println(Thread.currentThread().getName() + " with chunk: " + currentChunk + " finished!");
return null;
};
threads.add(thread);
}
try {
List<Future<Void>> futures = executorService.invokeAll(threads);
System.out.println("All threads finished: " + index.get());
return index.get();
} catch (InterruptedException e) {
e.printStackTrace();
return -1;
}
}
...
private List<List<Fmd>> splitToChunks(List<Fmd> fmdSearchArray) {
int size = fmdSearchArray.size();
List<List<Fmd>> lists;
if (size > IDENTIFY_BOUNDARY) {
int chunks = size / IDENTIFY_BOUNDARY;
if (size % IDENTIFY_BOUNDARY > 0) {
chunks++;
}
lists = new ArrayList<>(chunks);
for (int i = 0; i < chunks; i++) {
if (i + 1 == chunks) {
lists.add(new ArrayList<>(fmdSearchArray.subList(i * IDENTIFY_BOUNDARY, size)));
break;
}
lists.add(new ArrayList<>(fmdSearchArray.subList(i * IDENTIFY_BOUNDARY, (i + 1) * IDENTIFY_BOUNDARY)));
}
} else {
lists = new ArrayList<>(1);
lists.add(fmdSearchArray);
}
return lists;
}
The problem with this code is that it runs once. But at another try it doesnt come back from the native code of Caprture call.
So my question is:
How i can overcome this and make it work from my java code?
Or at least what is the direction of the solution?

The root cause is that this Identify function holds on to at least two references per returned Candidate after pushing it to the result array. It should instead release the references after pushing, so its use of the (limited) local reference table remain constant. You should file a bug about that.
The simplest workaround for now is to cut your fmdArray into 250-sized chunks and call Identify for each chunk.

Detect scrollTop() when hovered (handheld device)

I am using jQuery's scrollTop() for a fixed menu:
function fixed_menu(){
if( $('window').width() < 770 )
{
var menu = $('.col-left.sidebar');
var offset = menu.offset();
var trigger = offset.top;
$(document).scroll(function(e){
if($('body').scrollTop() >= trigger){
menu.addClass('fixed');
} else if ($('body').scrollTop() < trigger){
menu.removeClass('fixed');
}
});
}
}
fixed_menu();
When I am testing on my own phone (android device, Moto G 2nd gen), the if statement still works while scrolling.
When I am testing on an iPad mini, the if statement only initiates when the hover is done.
How can I make this function work on certain iOS devices, while the hover is still ongoing?

The scrollTop() function is problematic in different browser. You can try with $('html, body').scrollTop() and $(window).scroll()
function fixed_menu(){
if( $('window').width() < 770 ){
var menu = $('.col-left.sidebar');
var offset = menu.offset();
var trigger = offset.top;
$(window).scroll(function(e){
if($('html, body').scrollTop() >= trigger){
menu.addClass('fixed');
} else if ($('html, body').scrollTop() < trigger){
menu.removeClass('fixed');
}
});
}
}
fixed_menu();

keyup not working on Chrome on Android

I am using bootstrap typeahead.
It depends on this jQuery code to work:
el.on('keyup', doSomething() )
On Chrome on Windows it works fine. On Chrome on Android it doesn't. The keyup event is never fired. The element to which it is bound definitely has the focus.
This appears to be a recent development.
Chrome 28.0.1500.64
Android 4.1.2 SGP321 Build/10.1.1.A.1.307
Thanks
--Justin Wyllie

I came across this same problem earlier today. How can android chrome not support these key events! I assume you've found a workaround by now, but here's a fix that I came up with for now.
function newKeyUpDown(originalFunction, eventType) {
return function() {
if ("ontouchstart" in document.documentElement) { // if it's a touch device, or test here specifically for android chrome
var $element = $(this), $input = null;
if (/input/i.test($element.prop('tagName')))
$input = $element;
else if ($('input', $element).size() > 0)
$input = $($('input', $element).get(0));
if ($input) {
var currentVal = $input.val(), checkInterval = null;
$input.focus(function(e) {
clearInterval(checkInterval);
checkInterval = setInterval(function() {
if ($input.val() != currentVal) {
var event = jQuery.Event(eventType);
currentVal = $input.val();
event.which = event.keyCode = (currentVal && currentVal.length > 0) ? currentVal.charCodeAt(currentVal.length - 1) : '';
$input.trigger(event);
}
}, 30);
});
$input.blur(function() {
clearInterval(checkInterval);
});
}
}
return originalFunction.apply(this, arguments);
}
}
$.fn.keyup = newKeyUpDown($.fn.keyup, 'keyup');
$.fn.keydown = newKeyUpDown($.fn.keydown, 'keydown');

Sorry to say this but keyup/keydown events do not work for chrome browser in android.
There are other people who have reported this issue(Here and Here) from last 1 year and its not fixed yet. so it's better for developers to avoid using these events till it gets fixed.