We are running chromium 83 on an embedded system and experience some random tab crashes.
Is it possible to directly reload a tab in chromium, if it crashes (without showing the "Aw snap!" page)?
We're currently trying to patch the source code to get it working and those were our approaches so far.
(both in sad_tab_helper.cc -> SadTabHelper::RenderProcessGone()
Approach 1:
if (SadTab::ShouldShow(status)) {
web_contents()->GetController().Reload(content::ReloadType::NORMAL, true);
}
Approach 2:
if (SadTab::ShouldShow(status)) {
content::RunOrPostTaskOnThread(
FROM_HERE,
content::BrowserThread::ID::UI,
base::BindOnce(
[](content::WebContents* contents) {
contents->GetController().Reload(content::ReloadType::NORMAL, true);
},
std::move(web_contents())));
}
Both changes finally lead to crash of the entire browser.
It seems that chromium tries to reload the page but as said, it then crashes. The log we get are:
[1663:1671:0321/090914.211931:VERBOSE1:network_delegate.cc(32)] NetworkDelegate::NotifyBeforeURLRequest: http://127.0.0.1/login
[1663:1671:0321/090919.082378:ERROR:broker_posix.cc(40)] Recvmsg error: Connection reset by peer (104)
After that the entire browser crashes. Is there a way to do what we want or are we on a dead end here?
The second approach is suboptimal, SadTabHelper::RenderProcessGone only runs on UI.
Initiating navigation while handling notifications from any WebContentsObserver (SadTabHelper is a WebContentsObserver) must be avoided. It leads to problems. The both approaches attempt to do this. I suppose using base::PostTask instead of content::RunOrPostTaskOnThread should help.
if (SadTab::ShouldShow(status)) {
base::PostTask(
FROM_HERE,
{content::BrowserThread::UI},
base::BindOnce(
[](content::WebContents* contents) {
contents->GetController().Reload(content::ReloadType::NORMAL, true);
},
web_contents()));
}
My reputation is not good enough to comment so I can't leave this comment where it should be.
If you come across this particular solution, the required includes are:
#include "base/task_scheduler/post_task.h"
#include "content/public/browser/browser_thread.h"
#include "base/bind.h"
Related
I'm working on a Unity project for Hololens, that uses the camera to capture pictures, send them to a photo recognition API and displays the result. The project works perfectly fine in unity, but not on the emulator/Hololens.
Unfortunately, I wrote a lot of code at once, so i don't know at what point this problem started. The problem show's up after building the project and running it on the Hololens/emulator in Debug mode. On the Hololens, I see the starting window (the one you see after you open any application). After i place it, i see End showing splash screen. on the Output window in Visual Studio, and it just doesn't go any further (but doesn't freeze either, just does nothing).
I don't know where it's coming from, since no exceptions are thrown, but i suspect the camera is the cause. Earlier, i had to comment this line of code:
transform.position = Camera.main.ScreenToWorldPoint(new Vector3((CameraManager.Resolution.width * .5f), (CameraManager.Resolution.height * .5f), 10));
because the function ScreenToWorldPoint was throwing the following exception:
Screen position out of view frustum (screen pos 0.000000, 0.000000, 10.000000) (Camera rect 0 0 0 0)
As you see it says that the Camera rect's size is 0. I even tried directly logging the camera's dimensions to make sure (Debug.Log(Camera.main.pixelWidth + ", " + Camera.main.pixelHeight)), and sure enough, they were (0, 0) on the Hololens/emulator.
I made sure that webcam is supported, and that my camera settings are all set, but that didn't help either.
So i'm not sure if that's the cause of the problem or simply a symptom. And I can't start anywhere since neither the Output nor the Error window show anything wrong. Any help or suggestions would be greatly appreciated.
Thanks for reading!
Edit: Here's the entire output log from beginning to end.
Edit2: I don't know if this is significant, but if I paused execution (in Visual Studio), it always seem to be at Build/ProjectName/App.css => Line 78:
[MTAThread]
static void Main(string[] args)
{
var app = new App();
CoreApplication.Run(app); //<===== Here
}
You might want to check any of your Start() methods. You might have some code that is CPU intensive. Even if it runs smoothly in Unity, doesn't mean it will run easily on HoloLens since their CPU is not powerful.
Also, to avoid any Camera problems, make sur to use the Camera prefab from this
repository :
https://github.com/Microsoft/MixedRealityToolkit-Unity
Those are just some thoughts, hope it helps!
Turns out i didn't enable "Virtual Reality Supported" under Other settings in PlayerSettings. It's really dumb, but i hope this helps someone.
On the new iOS 11 Safari and MacOS High Sierra Safari, that trick of seeing if window.localStorage.setItem('test', 1); (see https://stackoverflow.com/a/17741714/1330341) throws an error no longer works, because it no longer throws an error, and it also properly sets the localStorage item. Has anyone figured out any other way to check for private browsing mode in the new versions of Safari?
Haven't actually tried it, but from reading Apple's document:
https://support.apple.com/kb/ph21413?locale=en_US
It lists various characteristics of private mode browsing (~snip):
When you use a Private Browsing window:
Each tab in the window is isolated from the others, so websites you
view in one tab can’t track your browsing in other tabs.
Safari doesn’t remember the webpages you visit or your AutoFill
information.
Safari doesn’t store your open webpages in iCloud, so they aren’t
shown when you view all your open tabs from other devices.
Your recent searches aren’t included in the results list when you use
the Smart Search field.
Items you download aren’t included in the downloads list. (The items
do remain on your computer.)
If you use Handoff, Private Browsing windows are not passed to your
iOS devices or other Mac computers.
Safari doesn’t remember changes to your cookies or other website
data. Safari also asks websites and others who provide those sites
with content (including advertisers) not to keep track of your
browsing, although it is up to the websites to honor this request.
Plug-ins that support Private Browsing stop storing cookies and other
tracking information.
From the above, in particular I found interesting that Safari specifically asks websites to "not track" the browsing. This could potentially be a mechanism to look for, to determine if using private browsing.
See this answer for an example:
Implementing Do not track in asp.net mvc
Again, haven't tested and unsure if it will work, but if not the list provides other potential options. HTH.
I find a solution here:
https://gist.github.com/cou929/7973956#gistcomment-2272103
var isPrivate = false;
try {
window.openDatabase(null, null, null, null);
} catch (_) {
isPrivate = true;
}
alert((isPrivate ? 'You\'re' : 'You aren\'t') + ' in private browsing mode');
Hope it helps :)
Quote from apple's website. https://support.apple.com/kb/ph21413?locale=en_US
Websites can’t modify information stored on your device, so services
normally available at such sites may work differently until you turn
off Private Browsing
So, store a test variable, change its value, then read the test variable.
If you get an exception, are unable to find the variable, the value didn't change or you get a null/undefined value back, they are most likely in private mode.
Alternatively, in private browsing, you have no stored search history accessible. So, redirect to a new page in your site on startup and then test if you have any previous history. If not and the fact that you are getting a Do Not Track most likely means your in private mode on safari.
Please note that I have not tested this. This is based off the information provided by Apple in the above link.
Thing that I realized is Safari throws a "Quota Exceeded" error in the Private mode. So here is what I did!
isPrivateMode: function () {
if (localStorage.length === 0) {
try {
localStorage.setItem('isLocalStorageAvaialble', 'That is being tested!');
localStorage.removeItem('isLocalStorageAvaialble');
return false;
} catch (e) {
return true;
}
}
}
Checking the length of localStorage is important for the fact that, if you are trying this method on a browser that supports localStorage, but is full, you still will get the "Quota Exceeded" error.
In private mode, the length of localStorage is always 0.
Hope this helps!
I've started a new React Native project and I keep getting the following warning:
Remote debugger is in a background tab which may cause apps to perform slowly. Fix this by foregrounding the tab (or opening it in a separate window).
It's a bit annoying so I wanna know how I can get rid of it? I'm running the debugger in Chrome and I moved it to a seperate window but it did not help.
If you have the Maintain Priority checkbox in the debugger window, try enabling it before you jump to any of the solutions below.
To get rid of the warning in your whole project add the following to your outermost Javascript file (most of the time that's index.js for React Native)
for react-native v0.63+:
Use LogBox:
https://reactnative.dev/docs/debugging#logbox
LogBox.ignoreLogs(['Remote debugger']);
for react-native v0.57 - v0.62:
import { YellowBox } from 'react-native';
YellowBox.ignoreWarnings(['Remote debugger']);
Reference this from the official React Native docs:
https://facebook.github.io/react-native/docs/debugging.html
react-native v0.56 or below:
Add the following early on in your code:
console.ignoredYellowBox = ['Remote debugger'];
Easy, simple and specific to that error. Works for me. Can substitute for any text you want.
this solution is work for me
open/move http://localhost:8081/debugger-ui (default path for remote debugging) on the separate window
maybe that could help :)
You can use React Native Debugger available at https://github.com/jhen0409/react-native-debugger It is a standalone app for debugging React Native apps during development.
Move http://localhost:*****/debugger-ui on the separate window.
Restart Remote JS Debugging.
For me warning went away by checking Maintain Priority Checkbox!
It is because of number of tabs are opened in the browser with React Native Remote Debugger UI tab. I also faced the same issue.
To overcome this warning message you can use any one method from the following:
Open an incognito tab then paste http://localhost:8081/debugger-ui on address bar and press ENTER. Finally reload the app (Command+R).
Close all the tabs in the browser. Keep only 1 tab opened then hit http://locahost:8081/debugger-ui then reload the app (Command+R).
As mentioned by #jakeforaker in one of the comment. The warning went away by simply opening the remote debugger in a separate window instead of a tab in your existing window of your browser (you have to reload your simulator though).
As the warning is saying keeping the remote debugger in the same window as other tabs
may cause apps to perform slowly
So i think simply suppressing warning as mentioned by #kjonsson:- console.ignoredYellowBox = ['Remote debugger']; doesnt seem to be best solution.
Since this commit in March 2017, you can enable the Maintain Priority checkbox. When enabled, it silently plays a base64-encoded .wav file to prevent the debugger's browser tab from entering low-power mode, which can affect websocket performance. This will effectively prevent the warning you describe.
This issue was resolved when I closed all open Chrome windows and started the Remove Debugging again. I had previously had open Chrome windows, so it 'seems' that having them open kills performance.
I think the accepted answer is no longer accurate (at least for React Native v0.57+).
The correct code is now:
import { YellowBox } from 'react-native';
YellowBox.ignoreWarnings(['Remote debugger']);
Reference this from the official React Native docs:
https://facebook.github.io/react-native/docs/debugging.html
I am on Macbook. I fixed this issue by bringing the Debugger window on main desktop, rather than on having it on separate desktop which it thinks is in "Background".
I had the same issue pop up yesterday. Googling it led to this Stack Overflow post. In one of the response (by adriansprod), he suggested:
Chrome debugger in it's own window fixes. But annoying problem
It is likely that your React Native debugger is not in its own Chrome browser window but in a Chrome browser tab. Pulling it out as its own window, as adriansprod suggest, fixed this for me.
The (very annoying) error message is handled by debuggerWorker.js, which sadly doesn't include any configuration options to turn off the message. So for the time being there are no ways you can configure your application to disable the message.
The related code is outlined below (original licence applies):
var visibilityState;
var showVisibilityWarning = (function() {
var hasWarned = false;
return function() {
// Wait until `YellowBox` gets initialized before displaying the warning.
if (hasWarned || console.warn.toString().includes('[native code]')) {
return;
}
hasWarned = true;
console.warn(
'Remote debugger is in a background tab which may cause apps to ' +
'perform slowly. Fix this by foregrounding the tab (or opening it in ' +
'a separate window).'
);
};
})();
As you see, no configuration options are used, the whole thing is scoped off locally (see the above repo link for further details).
I also have faced with same issue about one week ago and finally i have found solution that works excelent for me
It called reactotron, you can find it here - https://github.com/reactotron/reactotron and you can use it to:
* view your application state
* show API requests & responses
* perform quick performance benchmarks
* subscribe to parts of your application state
* display messages similar to console.log
* track global errors with source-mapped stack traces including saga stack traces!
* dispatch actions like a government-run mind control experiment
* hot swap your app's state
* track your sagas
I hope my post was helpful and you never will faced with this tedious warning .
Good luck
I use this in index.js
if (__DEV__) {
console.ignoredYellowBox = [
'Remote debugger',
'Warning: isMounted… is deprecated',
'Module RCTImageLoader'
];
}
I had minimised the "http://localhost:8081/debugger-ui/" window. Just opening it up (un minimising), and reloading the app removed the warning.
there might be chances that Another debugger is already connected to packager.
so close your terminal and debugger google chrome.
if you are using visual studio's package manger then don't start package manager by Mac/other os terminal command.
so close all terminal and stop on going package manger and google chrome debugger.
start the process again.
Similar to Akshay Vijay Jain, mine went away by ticking this box!
I've got the following code in a thread in my application:
while (true) {
if (ready) {
progressIndicatorController.value++;
return;
}
}
The ready variable is changed from a delegate method. This code works great when I open the application by clicking the "Run" button in Xcode's toolbar. However, if I open this application's .app (which I create by clicking Product > Archive and then following the steps) this code somehow doesn't work anymore.
progressIndicatorController.value is never incremented and this if-statement never evaluates to true. What could cause this problem?
This is probably caused by optimization from the compiler.
When you build with Archive, XCode enabled optimization in the compiler that could throw this kind of code away. I think setting the ready variable to volatile could fix your problem, altough if I were you I'd just try to rewrite it so it doesn't trigger this problem.
You can test with optimization turned on by choosing Edit Schemes in the scheme dropdown. Then set Build Configuration to Release in the Run MyApp.app. Don't forget to set it back to Debug when you're done though, as the debugger gets somewhat confused when optimization are on (i.e. you can't see the value of most variables, some breakpoints may behave erratically, etc...)
It is quite tricky because I wanna to take the result as a sign for later process. If a window flashes, it would be weird to user.
(This is not an answer, just refreshing the question and adding some details:)
Suppose you need to open a video capture driver (webcam) with code like this (delphi, but easy understandable):
result:= SendMessage(hCapWnd, WM_CAP_DRIVER_CONNECT, FDriverNo, 0);
It works fine, except when camera/webcam is busy (opened by another application, for example moviemaker). Then the capture driver suddenly shows a "select video source" window that blocks your application (but not moviemaker, which keeps recording). SendMessage function becomes modal and will not return until undesired window is closed.
Is there a way to detect if a capture driver is busy with another application before connecting to it?
Thanks
This function return number of available webcams and scanners on your mashine(с++).
int GetNumCam()
{
int MemberIndex = 0;
HDEVINFO dev;
SP_DEVINFO_DATA DeviceInfoData;
DeviceInfoData.cbSize = sizeof(DeviceInfoData);
dev = SetupDiGetClassDevsA(&GUID_DEVINTERFACE_IMAGE, NULL,NULL,DIGCF_PRESENT);
if (dev == NULL)
{
MessageBoxA(0,"it is null..","vse propalo",0);
return -1;
}
else
{
while(SetupDiEnumDeviceInfo(dev,MemberIndex, &DeviceInfoData))
{
MemberIndex++;
}
}
return MemberIndex;
}
I wonder if sending a broadcast message could do. I mean, suppose you send a VFW info request message to all windows in system. Then, all windows actually doing VFW processing will answer their status info and you will know which of them are doing video processing; some kind of shouting "anybody here...?"
But it looks like brute force, or even hacking. May be I'll test it... may be not. I am sure there must be a smarter way to know if a video capture driver is busy with another application.
(btw... anybody here?)
I was afraid it would happen.
First I enumerated all opened windows in system, then executed this instruction for each HWND, which just asks for driver information (pascal syntax):
SendMessage(h, WM_CAP_DRIVER_GET_NAME, length(driver)*sizeof(char), LPARAM(PChar(driver))
According to Msdn help, WM_CAP_DRIVER_GET_NAME results are:
"Returns TRUE if successful or FALSE if the capture window is not connected to a capture driver"
(Bold is mine). So it appears a good way to know:
1-If the window is capturing. So it is applyable to ALL windows, don't matter if they are capturing or not.
2-And if so, it tells you which driver is using.
However, after first test round, the results were:
- Task Manager (it was running) crashed and closed
- Explorer crashed and closed (reopened again automatically)
- Belkin Wireless monitor (my pc's WiFi driver) crashed and closed
- Eset Nod32 antivirus crashed, did not closed
The first conclussion can be that this is a bad way to locate which applications are capturing video in a system. But I must ensure to dismiss the possibility of a bug.
I'll keep reporting.
This is not a good solution my any means but I found out that if you disable and enable the camera in device manager immediately before sending the WM_CAP_DRIVER_CONNECT message then for some reason the video source window does not appear and the function returns and works OK.
So for a script I was working on I had to download devcon.exe and I put "devcon disable *PID_3450*MI_00*" and "devcon enable *PID_3450*MI_00*" immediately before my WM_CAP_DRIVER_CONNECT and it seems to work consistently. Unfortunately you need to run as administrator to be able to disable/enable devices. Again not a proper fix by any means...
(Hardware ID of my cam is USB\VID_0AC8&PID_3450&MI_00)