I'd like not to be disturbed by the browser that is launched by Selenium RC. Is there any way how to force it to run the browser on another X11 desktop? Running Selenuim server on another desktop does not solve the problem.
Vít Šesták 'v6ak'
Yes, on Linux you can start for example Xvfb (X in a virtual framebuffer) to provide an X environment and launch the selenium controlled browser in that.
We are doing that for years on headless server boxes and that work just fine (well from time to time you have to shoot a hanging browser, but that is the same on a desktop).
First install Xvfb:
sudo apt-get install xvfb
Then start selenium server in background via xvfb:
xvfb-run java -jar ~/Downloads/selenium-server-standalone-2.45.0.jar
Finally run your tests as usual (phpunit, or whatever you are using).
You should be able to do this by setting the DISPLAY environment variable to a different display than the one your desktop is using. This must be set when starting up the Selenium server. Selenium will then use this value when starting up the browsers.
I solved this using workspaces/viewports not X using devilspie.
This addon lets you put a custom title on firefox, I installed it on my selenium profile:
https://addons.mozilla.org/en-US/firefox/addon/57/
I replaced Mozilla Firefox in the titlebar with FFSELENIUM
Then use devilspie to pin those windows to firefox:
http://ubuntu-tutorials.com/2007/07/25/how-to-set-default-workspace-size-and-window-effects-in-gnome/ http://foosel.org/linux/devilspie
Then create a file .devilspie/selenium.ds with the contents:
(if
(and
(contains (window_name) "FFSELENIUM")
)
(begin
(set_viewport 3)
)
)
Woot firefox windows in the third viewport!
Related
I can't debug my react native / expo app in WebStorm, but I can in Visual Studio Code.
I have a react native app that I can successfully run using expo. I set up the configuration according to the info on this site:
https://blog.jetbrains.com/webstorm/2018/02/webstorm-2018-1-eap-181-3263/#debugging-expo
When I click "run" or "debug" and "Debug JS Remotely" is turned off, the bundler builds a JS bundle and the app starts on my devices.
But once I turn on "Debug JS Remotely", the bundler only says:
Starting custom debugger by executing: : [my working directory] [my working directory again]
and never goes on to build the JS bundle.
After a short while, a red screen appears on the device, and it says:
Unable to connect with remote debugger
Timeout while connecting to remote debugger
onFailure
DevSupportManagerImpl.java:851
onFailure
WebsocketJavaScriptExecutor.java:83
run
WebsocketJavaScriptExecutor.java:142
handleCallback
Handler.java:790
dispatchMessage
Handler.java:99
...
The bundler doesn't say anything more than the message above.
As soon as I turn off remote debugging, the the bundler starts building again and finishes with
"Building JavaScript bundle: finished in ...ms"
as usual, but the device, again, shows the red screen with aforementioned messages, or doesn't show anything but the regular white screen with a message at the bottom saying "Downlading JavaScript bundle 100,00%". I either have to stop and restart within WebStorm or I have to close the expo app and re-open it. Either way will make the app start again on my device.
This is my first time working with javascript, react native, expo, so I am not sure, but it should be possible to debug in WebStorm directly, right? As mentioned, in Visual Studio Code everything including debugging works fine, so I don't think it's anything in the code. Maybe someone here has an idea where the problem lies?
node -v: v11.6.0
react-native -v: react-native-cli: 2.0.1, react-native: 0.57.1
Any help will be greatly appreciated! Thanks in advance!
PROOF
HOWTO
Looks like there's been an update to how you can setup Configurations since you tried.
Try changing the Bundler host to 127.0.0.1 and try again.
I just setup Webstorm v2020.2 yesterday to debug my Expo application and it is working as expected.
These instructions were all that I required: https://www.jetbrains.com/help/webstorm/react-native.html#ws_react_native_create_run_config
On the main menu, go to Run | Edit Configurations, click icons.general.add.svg and select React Native from the list. The Run/Debug Configuration: React Native opens.
Choose whether you want WebStorm to build and launch the application for you:
Select the Build and launch checkbox if you are launching your application for the first time or if you have updated its native code since the last run.
If your application uses Expo, clear the checkbox because this bundler takes care of the process itself.
If you selected the Build and launch checkbox, choose the target platform, Android or iOS. Depending on your choice, WebStorm will run the bundler with react-native run-ios or with react-native run-android.
Optionally, type the arguments to be passed to React Native, for example, specify the simulator type through the ‑‑simulator flag: ‑‑simulator="iPhone 4s".
In the Bundler host field, specify the host where the React Native bundler runs, the default value is localhost.
If you are using Expo, change the default bundler host to ensure successful connection with the debugger. This connection may fail because WebStorm by default uses localhost to start debugging while Expo expects 127.0.0.1 or an external IP address depending on what is selected in Connection field of the Metro bundler configuration. See Debugging a React Native application that uses Expo for details.
In the Bundler port field, specify the port on which the React Native bundler runs, by default 8081 is chosen, learn more from the React Native official website.
Choose the Node.js interpreter to use. This can be a local Node.js interpreter or a Node.js on Windows Subsystem for Linux.
Specify the path to react-native-cli and the working directory of the application. Optionally, type the environment variables for react-native run-android or react-native run-ios.
By default, WebStorm starts the React Native bundler automatically when you invoke the run/debug configuration. If you have already started the bundler from outside WebStorm, for example, from the command line, you can re-use it without stopping and restarting. Select your bundler in the Before Launch area and click Remove.
The issue is tracked at WEB-35719, please follow it for updates
I have a lettuce test suite using selenium and everything works just fine on linux.
After I installed django and everything that's needed on windows to test the suite on IE8,9 too, and I tried to run the test, it only opens my browser and says that the test passed with 0 features, 0 steps etc. The same test suite on linux runs just fine.
What do I need to make them work on windows 7 too ???
I use python for my test.
Maybe it's not the best solution for you problem, I'm not even sure that it's valid for IE, but you could try to launch IE remotely so you will not need to launch whole application on windows (i.e. if you were using gunicorn as wsgi server, you couldn't launch on windows at all).
Python-selenium has functionality to use remote browsers, I'm just not sure about IE.
I'm looking at Selenium Server at the moment, and I don't seem to notice a driver that supports headless browser testing.
Unless I'm mistaken, it doesn't support it. If you're on X, you can create a virtual framebuffer to hide the browser window, but that's not really a headless browser.
Can anyone enlighten me? Does Selenium support headless browser testing?
you need not use PhantomJS as an alternative to Selenium. Selenium includes a PhantomJS webdriver class, which rides on the GhostDriver platform. Simply install the PhantomJS binary to your machine. in python, you can then use:
from selenium import webdriver
dr = webdriver.PhantomJS()
and voila.
The WebDriver API has support for HTMLUnit as the browser for your testing. Ruby people have been using Capybara for a while for their headless selenium testing so it is definitely doable.
I know this is a old post. Thought it will help others who are looking for an answer.
You can install a full blown firefox in any linux distribution using XVFB. This makes sure your testing is performed in a real browser. Once you have a headless setup, you can use webdriver of your choice to connect and run testing.
Headless browsers are a bad idea. They get you some testing, but nothing like what a real user will see, and they mask lots of problems that only real browsers encounter. You're infinitely better off using a "headed" browser (i.e., anything but HTMLUnit) on a headless environment (e.g., Windows, or Linux with XVFB).
I notice that you say that using an X framebuffer isn't a true headless solution, however, for most, I think it would be acceptable. In addition to that, this service will help get that going for you if you are interested in that as a solution.
Selenium does support headless browser testing in a way. Docker Selenium is maintained by SeleniumHQ. Those docker containers come with xvfb support with them out of the box. There are tools like PhantomJS that you can hook up with Selenium. However, it's not officially supported by Selenium itself.
Much like what others have described, PhantomJS isn't really recommended. The whole point of Selenium is to automate browsers. But why automate a browser no one uses? I never understood how that was overlooked so often by developers..
Yes. Selenium support headless browser testing and it's more faster as well as convient for big amount of test-cases execution.
ChromeOptions cromeOptions = new ChromeOptions();
//Location of browser binary/.exe file
cromeOptions.setBinary("/usr/bin/google-chrome-stable");
cromeOptions.addArguments("--headless");
cromeOptions.addArguments("--no-sandbox");
cromeOptions.addArguments("--disable-gpu");
cromeOptions.addArguments("--window-size=1920,1080");
WebDriver webDriver = new ChromeDriver(cromeOptions);
Yes ,selenium supports headless browser testing...but i found HTMLUnit failing most times...I was searching for an alternative...PhantomJs was really good.you can definitely give it a try it was very fast when compared to other browsers...It is really good for smoke testing...
http://phantomjs.org/
With ruby and macOS: brew install phantomjs then:
driver = Selenium::WebDriver.for :phantomjs
Yes Selenium supports headless browser testing.Headless browsers are faster than real time browsers.
Install chromeDriver and google-chrome-stable version on the linux server, where the tests will be triggered and add the same binaries in your code.
code snippet:
private static String driverPath = "/usr/bin/chromedriver";
static
{
System.setProperty("webdriver.chrome.driver", driverPath);
options = new ChromeOptions();
options.setBinary("/usr/bin/google-chrome-stable");
options.addArguments("headless");
driver = new ChromeDriver(options);
}
Here's a "modern answer" on how to use Selenium with xvfb and Firefox driver in an Ubuntu Linux environment running Django/Python:
# install xvfb and Firefox driver
sudo su
apt-get install -y xvfb firefox
wget https://github.com/mozilla/geckodriver/releases/download/v0.19.1/geckodriver-v0.19.1-linux64.tar.gz
tar -x geckodriver -zf geckodriver-v0.19.1-linux64.tar.gz -O >
/usr/bin/geckodriver
chmod +x /usr/bin/geckodriver
# install pip modules
pip install selenium
pip install PyVirtualDisplay
You can then follow the Django LiveServerTestCase instructions.
To use the driver you just installed, do something like this:
from pyvirtualdisplay import Display
from selenium.webdriver.firefox.webdriver import WebDriver
driver = WebDriver(executable_path='/usr/bin/geckodriver')
display = Display(visible=0, size=(800, 600)).start()
# add your testing classes here...
driver.quit()
display.stop()
My complete working environment is Linux based (Ubuntu for desktop and server). I use Hudson and Selenium to execute tests on my EE6/ZK web application with any browser available for Ubuntu.
But how would I test my application with IE (Windows) without setting up a Selenium server in Windows. Thought I could run Windows in a VMWare or Virtualbox, but still it needs a "local" selenium and how would I trigger it ? Install hudson as well and pretty much replicate the complete CI in Windows ? How is MAC ?
Any tips or experience with cross-browser-platform-OS automated testing ?
Sven
You will need to use Selenium Grid. There is a Selenium plug-ins for Hudson available, and this article has a guide how to setup Continuous Integration with Hudson and Selenium Grid. And yes, you will have to install Windows in a virtual machine.
I develop web applications using Django on Mac OSX 10.6. I use Django's built in local development server which I run on my computer's IP (such as: http://192.168.0.11:8001/). I test my applications in Firefox, Safari and Chrome and all display fine. I use Sun's VirtualBox with 3 different instances of Windows XP that have IE6, IE7 and IE8 on them.
For whatever reason, these sometimes just don't display the Django sites. They come up with 'The page cannot be displayed'. Eight times out of ten, they display fine and function normally but for no reason at all they won't display. Sometimes restarting Django's local development server from the Terminal will fix the problem, sometimes it won't.
Is there some sort of VirtualBox settings or Django settings that I need to set to ensure smooth operation of this? Am I overlooking something? Has anyone else had these problems?
By default the dev server binds to localhost. to run it on all interfaces use ./manage.py runserver 0.0.0.0:8000
I found the best way to do this is using the normal localhost:8000 or localhost:8001 (any port you need, especially if running multiple Django projects at the same time) in the Mac Terminal.
Then in VirtualBox with Windows, using the ipconfig command in the Command Prompt to find out the IP settings. Grab the Default Gateway (mine was 10.0.2.2 for example) then, in the browser, use this as the address to get to your Django project along with the original port setting that was set in the terminal. So in this case: http://10.0.2.2:8000/
I've been running it like this now for almost a week and it seems to of stopped the problems I was getting before totally. I couldn't really tell you why though!