I have a code, which captures a given window by PrintWindow function. Specifically I do capture web-pages in a browser. This code is tested in a standalone Windows application, and it works ok. The same code is incorporated into NPAPI plugin and loaded into Google Chrome. There problems happen. If Chrome has only one tab, the code works OK. If Chrome has 2 tabs, the code returns black box of expected size. I tested the standalone application and the plugin at the same moment, on the same window handle (Chrome_RenderWidgetHostHWND). Application does ALWAYS work, the plugin does ALWAYS fail if multiple tabs are open in the browser.
I'd really appreciate, if someone give a clue how to fix this.
Thanks in advance.
NB. Other methods of capturing windows are inapplicable due to even greater shortcomings for my usecase.
I bet you're running into sandboxing issues.
Your app is likely running at the "normal" integrity level, so it has no problem posting the WM_PRINT or WM_PRINTCLIENT message to the Chrome process. (PrintWindow uses these messages under the covers.)
When you're a plugin in Chrome, I believe your code runs as a "low" integrity process (which reduces the chances of things like shatter attacks). Thus it doesn't surprise me that sending the message to a window in another process fails.
Unfortunately, that doesn't explain why it works when Chrome has only one tab. Are you finding the same instance of Chrome_RenderWidgetHostHWND in both the app and the plugin in the case where you have multiple tabs open? In general, there's one of these windows per tab. How are you finding the window?
Related
I'm using opengl es 3.0 API with the android studio ndk to create apps.
But I've encountered a very huge problem. I've created a demo app, all it does it change the background color of the screen from white to black and vice versa, every frame. And so when I go to minimize this app, I still see it rendering the background, mostly at the edges of the screen, and not in full color but still very strongly apparent. And it doesn't go away when I close the app, when I restart the device, or when I run "kill apps" on it. Only a factory data reset fixes the issue, so it's not easy for me to debug this.
This is the relevant code that I'm using for when the app is minimized and receives the APP_CMD_TERMINATE event:
eglMakeCurrent(engine->display,EGL_NO_SURFACE,EGL_NO_SURFACE,EGL_NO_CONTEXT);
eglDestroySurface(engine->display,engine->surface);
engine->display = EGL_NO_DISPLAY;
engine->surface = EGL_NO_SURFACE;
I've error checked that eglDestroySurface() is successful.
And I've put debugging messages in to make sure that the main draw loop is NOT executing when the app is minimized. But the problem persists and I don't know what to do about it. Thanks for any help.
UPDATE: well, no one has responded, and I still don't know what to do. Could it be related to threads?
UPDATE: Still can't determine what it is, but for some reason it's messing with the System UI. Willing to upload my entire source code somewhere if someone would be willing to go through this with me, as I'd really like to be able to continue working on my game engine.
Is it the "Strict Mode" developer option on the device settings, perhaps?
That one flashes the screen if an app is blocking.
It would explain why a factory reset changes behaviour.
The answer is not a solution here. The above comment by the user columbo was correct.
I've demoed switching from black to white at high framerates on 3 different android devices, and also my Linux Desktop, all via the openGL api, and it has exhibited this issue on all the devices. So what he said must be correct: this is a problem with LCD monitor technology itself. Interestingly, doing completely random colors does not cause this problem.
While we all got tangled up in this topic, it came to light that, while UWP apps do have top level windows and you can send (sensible) messages to them (*), and these do get through and have the expected effect, Spy++ doesn't ever seem to see those messages, or indeed any others that Windows itself might send.
Can anyone shed any light on this?
Just to provide a bit of background, Spy has no problem locating and inspecting these windows (the 'Finder Tool' works just fine) but message logging is totally mute. I have tested the Settings App, the On Screen Keyboard, Calendar and Calculator, and I did tick all the appropriate 'Additional windows' checkboxes in Spy. I also used Process Explorer to check that Spy runs at 'High Integrity Level' (because it runs elevated, presumably), and it does.
(*) Yes, I know we're not supposed to but, well, you know.
Um, embarrassing. You need to use the 64 bit version of Spy to spy on a 64 bit app (and vice-versa). Then it works fine and you get what you expect.
Of course, it would be nice if Spy told you when you try to trace messages in an app of the wrong bitness but it doesn't, so please be warned.
I have an application with web interface. Unfortunately, it has all disadvantages of being a web page:
It doesn't have a standalone window, so users cannot manage it via the taskbar.
Users see the address line with something like 'http://localhost:8080' that is not a good idea for home users.
If users click on a tray icon, there is no way to activate the tab in a browser, which contain the application interface.
So, it would be nice to have a wrapper application with a browser within.
In case of IE I know it's possible to create a window with Trident ActiveX component. But what if it's Windows XP with IE6 but installed latest Chrome? I'd like to prefer Chrome since it supports a lot more features which the user will never see.
So, is there a way to wrap a page into Chrome/Firefox and make it look like a standalone application, if one of them is presented in the user's system? (The application shouldn't install anything large, so Chromium build is not an option).
P.S. I'm not interested in supporting other platforms than Windows.
Regards,
Take a look at Chrome Apps.
I hope helps you.
I've got a Selenium test which crashes my browser (Firefox) whenever it runs. It's just a simple test (login, logout, login, logout, login ... crash) and I can't reproduce the crash when I do the same things manually, so I think the problem is with Selenium itself.
However, the issue could be that my Firefox starts with different options than when Selenium starts it. I could confirm that's not the case if I could start Firefox the exact same way Selenium starts it (eg. without plug-ins, with all history cleared, etc.) ... but I'm not sure how to do that.
Can anyone tell me which options to use to make Firefox run the same way as when Selenium runs it?
I was never able to find out how to emulate Selenium's use of Firefox. I tried using a fresh Firefox profile, but that still seemed to have differences from how Selenium ran Firefox.
Instead I wound up re-writing my test to be two tests, and then it was able to log in/out twice in the first test and once in the second test, avoiding the issue I was trying to debug (which only manifests when I try to log in/out 3 times in the same test).
If anyone can provide a better explanation of how to emulate Selenium's running of Firefox I'd be happy to accept that answer.
Is it possible to lock a user in chrome using .dll files (NPAPI plugin) or any other method?
I want to invoke chrome browser in highly controlled environment preferably on Windows. I would download chrome for businesses and write policies such that user isn't allowed to access any url except one.
I would create a packaged app and create an application shortcut on desktop for user which goes to the url. So it will open in KIOSK mode. But I would want that user shouldn't be able to navigate away from chrome, specifically that window and shouldn't be able to use any other program until he logs out of the application or explicitly closes the window. That means disabling function /ctrl/alt and window keys.
Something like this but with chrome.
Even opening chrome itself in 'locked' mode is not a problem.
Has someone ever implemented it? Is it possible using NPAPI plugin or any other method? If anybody could direct me towards any such resources I would be grateful.
No, this is not realistically possible using an NPAPI plugin, at least not by itself. At minimum you'd need an extension (see http://npapi.com/extensions).
If you loaded it from an extension it might be possible to get the window handle of the browser and such, but I think you'd probably be better off with a separate app.