I am trying to write some integration tests in rspec/capybara/selenium for my Rails 5 app. I have recently started testing some javascript features in the app with headless Chrome.
I struck a problem where I was unable to choose a select element on the page with capybara despite this working fine when I loaded the site manually in chrome in development environment. After some investigating I figured out that the select element was not currently visible. It is hidden when the page loads but should then be made visible immediately on loading by my javascript.
I disabled headless and paused my test with a quick and dirty sleep 60. I then looked at the javascript file in Chrome's developer tools and discovered that it has loaded an old version of the file with none of my recent changes. The file it has loaded no longer exists in my app so it must be being cached somewhere. Any ideas how this might be occurring and how I can fix it?
Related
I am writing capybara tests. There is a link I have in the view. When I click over the link that links open a pop-up js warning. I have configured Js. in capybara by using phantomjs and petergiest gem.
Without the requested information it's impossible to give an exact answer, but the error you are seeing means the app is requesting a non-JS response (probably HTML). This could be occurring for a couple of reasons
You're not actually running the test with a JS supporting driver. I don't see any js metadata on your scenarios so depending on how you've configured Capybara/RSpec this could by your issue. To confirm, swap from Poltergeist to using Selenium with Chrome or Firefox (non-headless while trying to debug) so you can see if the browser actually starts
You have a JS error preventing JS from running so a normal request is being made instead of XHR. This could be because you actually have a bug in your JS or because you're using Poltergeist/PhantomJS which is massively out of date in JS/CSS support. To test this, swap to using Selenium with Chrome or Firefox and look in the developer console.
Your link isn't correctly configured to make an ajax request - This is impossible to tell without the HTML of the link
Additionally, neither of the tests shown in your image are actually asserting/expecting anything so it's very unclear what exactly you're trying to test.
I'm running Firefox 36.0.4 on Windows 7 32-bit. I've diabled all add-ons, extensions and user scripts before retesting this.
I'd like to step through JavaScript code that is served up in a <script> tag in the HTML document being produced by a Java (Tomcat) web server.
Unfortunately, when I select the HTML document under Debugger > Sources, the source of the page returns to the login page of the application - it appears that session information is not being used to request the source.
I stepped through the server-side code and found that the correct session cookie values were being sent for the real page request and some AJAX requests sent by the page. However, when I tried to load the page source in the JavaScript debugger, I found that an incorrect session cookie was being sent by the JavaScript debugger.
I can replicate this behaviour in other webapps, not just my own. For example, Stack Overflow:
Is this a configuration issue, or a bug in the Firefox Developer Tools?
I can't reproduce your problem using StackOverflow as an example, at least in Firefox Developer Edition ( currently version 38 ):
One thing that might help - try disabling the cache while the toolbox is open - this setting is in the developer tools setting panel ( click on the 'gear' icon at the top right of the toolbox ):
After reviewing canuckistani's answer, I downloaded Firefox Developer Edition. Seemingly, the problem was fixed.
Five minutes in, I became sick of being asked whether to remember passwords and having to manually clear session cookies (I prefer being able to do it by simply closing the browser) - it makes testing easier.
As per usual, I went to Options > Privacy > History to disable this behaviour, by setting the value to Never remember history.
Changing this setting requires the browser to restart. However, upon restarting, I once again saw the same erroneous behaviour - the wrong session cookie was being sent to the web application again.
The workaround here is to not use the Never remember history setting. I have filed a bug report at Mozilla.org Bugzilla.
In my rails project sometimes chrome doesn't stop showing "loading circle icon" even after finished a loading. At least it appears all HTML elements are shown and JS works fine.
It doesn't happen in local development. Neither with Firefox and Safari in server environment.
It happens only with chrome in server environments.
What should I do to determine the cause of the problem?
While debugging a JavaScript program I was writing with Firebug, the Script panel stopped working and I got the following message:
I went through all menus and sub-menus I could find but didn't see a option to re-enable this feature. The console.log() command doesn't work in Firebug as well (while it does work in the built-in devtools' console).
How can I re-activate the Firebug Script panel? What may have caused this?
This may have different causes. One of them is described in issue 5646, which is related to going back and forward in the browser history.
In such cases it normally requires a browser restart to work again.
Note that Firebug up to version 1.12 is based on an old debugger API exposed by Firefox called JSD, which is unmaintained and buggy. Since version 2.0 Firebug uses a new debugger API, which fixes this problem.
Coincidentally, I just "fixed" the same problem with my Firebug.
I went through re-installs, options resets and such and nothing helped.
After some investigation it boiled down to the content of the page I was debugging.
I have a function called by jQuery when page load is finished:
<script>
$(onPageLoad);
</script>
Now, this function asks user to enter his name using prompt() function. It appeared that Firebug wouldn't activate script panel until prompt() is closed. That is not a problem when you have just one page opened since you just close the dialog and everything works. However, it is a problem when you have two same pages opened. Script panel is non-operational until you close both prompts.
I was facing this problem recently. I found that my Firebug was not getting updated because my Firefox was stuck on 29.x. Once I updated my Firefox the new Firebug got installed. It supports only Firefox 30+.
Note: I was using OSX v1.9.3.
I was having same problem, and, thanks to Roman Guralink, I figured out that the the dialogue Firefox prompts you to insert the master password was the cause. When you set a master password for all password stored, FFox usually opens a dialogue to ask it just after program started. BUT, sometimes this dialogu goes behind the main window, and you don't notice it is there.
I would be grateful for some pointers on how to troubleshoot this problem. Quick summary of the situation:
I have a large document (200K) divided into five main sections of inequal length, each of which corresponds to a jQuery UI tab.
The page works perfectly in Safari for Windows, Safari for Macintosh, Opera, Firefox, Chrome, and IE.
The page used to work on the iPad too but it no longer does, and it's the iPad that has changed, not the page.
On the iPad, the tab-content is now getting cut off. E.g. one of the pages is a glossary that runs from A-Z and the page cuts off at the letter -H-. It used to scroll all the way to -Z-. For some reason, iPad Safari is not giving each Tab the full amount of vertical space it needs for its content.
I've looked at the jQuery UI code for the show tabs and it appears to be changing CSS classes hide/show, but I'm no ninja javascript coder. How do I begin to figure out what's wrong on the iPad when my page works as expected on every other browser?
EDIT: The page seems to be working fine on the iPad2. It could be a caching issue and the page might stop working at some point even on the iPad2--I could only test at the Apple store. But I believe the markup and coding and jQuery ui are essentially OK, and it's a iPad Safari issue.
I can't trouble-shoot myself without the code, but the by far best way to trouble-shoot situations like this is to use a tool called Weinre. Obviously if one browser (the iPad's) is having trouble, you need to troubleshoot directly with it. That's difficult since the iPad doesn't have developer tools, but Weinre can actually give you (most of) that, over the network.
Follow the instructions here, but at the least you need to provide a ~/.weinre/server.properties that contains this:
boundHost: -all-
httpPort: 8081
reuseAddr: true
readTimeout: 1
deathTimeout: 5
That will tell weinre to listen on every IP on the port 8081. Start Weinre (via the OS X runner or with java -jar weinre.jar on the command line.
Then you add a special script tag in the main page:
<script src="http://YOUR_IP_NUMBER:8081/target/target-script-min.js"></script>
After this you start Safari or Chrome and go to http://localhost:8081/client/. If everything went as planned you will see the Weinre interface, which is a subset of the WebKit developer tools.
Now connect to your development machine with the iPad or simulator. If the script tag is correct Weinre makes a connection to the iPad and you have a fairly large subset of WebKit's developer tools at your disposal for trouble shooting. Good luck!
I'm sure this isn't the answer you're hoping for, but it sounds like a bug that should be reported to Apple.
You have a page that works perfectly on every other browser and I think its in Apple's best interest that it works on the iPad as well.
There might be other pages out there that used to work, but now don't, so I would consider this bug to be pretty high priority.