I would like to test that my code does not exhibit a race condition where css files are served slowly. Is there a tool that can help me test this?
Fiddler has "simulate modem speeds", but I can't specify that only css files should be delayed.
Any other quick suggestions? (I don't want to spend too much time on setting this up, this should be a drop in tool that just does the job).
You can disable CSS styles with the Web Developer Toolbar (there are seperate Chrome and Firefox downloads).
If this isn't acceptable you could reference this StackOverflow question to put some JavaScript on the page:
How to load up CSS files using Javascript?
and wrap it in a
setTimeout(function() { // code here }, millisecondsDelay);
to delay the loading however long you'd like.
Fiddler Delayed Responses Extension seems to be what I want.
Related
At the end of my tests Capybara automatically navigates to "about:blank" in order to set up the next test. Sometimes the application I'm testing will throw a popup alert if the user leaves the page (which is expected). I have some code to handle this:
begin
page.driver.browser.navigate.to("about:blank")
page.driver.browser.switch_to.alert.accept
rescue Selenium::WebDriver::Error::NoAlertPresentError
# No alert was present. Don't need to do anything
end
This works fine on Firefox, Chrome, and IE. But for some reason on Safari the navigate command hangs, I assume because of the popup. Anyone know a workaround for this?
There is no simple workaround for this at this time in any version of Selenium language bindings. It is a known issue the Selenium team is not interested in resolving. Fundamentally, it is due to the architecture of Safari and consequently the architecture of the Safari Driver.
The JavaScript of the Safari Driver extension does not know about most of the alerts and popups and dialogs that appear as modal Cocoa layer windows.
It also cannot interact with them.
There is a way but it won't be easy and nobody's done it.
You would need to use Cocoa.
So you would want to use RubyCocoa in this case.
(or PyObjC if you used Python)
You would then possibly also want a sidecar app actually written in Objective-C.
The trick would be to use the AX (Accessibility API) and a separate process to observe if there is an alert as the front window and poke at its labels and buttons' text as visible to the AX APIs.
AX APIs are probably exposed in RubyCocoa via the ScriptingBridge.
However, you would need to add your 'app' to the Security preference pane's list of things allowed to control the computer.
With that, you could detect the window and handle it.
It could be fairly brittle across web sites, but if built well, you could handle expected conditions.
You could try to confirm like this which I believe should work across browsers
# click ok to confirm
page.evaluate_script('window.confirm = function() { return true; }')
I'm using PreloadJS as part of an application that's being built with the CreateJS libraries. PreloadJS is being used to pull in graphics and audio files listed in a manifest. I've setup a progress bar and hooked the preloader's fileprogress and fileload events to update it. I'm getting progress updates as the images load, so I can see the progress bar crawling along, but I never get fileprogress updates for the audio files, just the fileload (file is completely loaded) so the progress bar sits idle and then suddenly jumps to 100%. Since the audio files are by far the largest assets, this creates something of a problem as far as providing meaningful load progress to the user.
I've looked at the documentation for PreloadJS and it indicates that XHR loading is the preferred method because it does provide the progress updates, but that PreloadJS can fall back on things like tag-based (<audio>) loading of audio files.
None of the file loading is local or cross domain, so it would seem to me that PreloadJS ought to be using XHR.
Is there a way to force PreloadJS to use XHR for everything so I can get consistent progress updates or why would I not be getting progress events for these much larger files?
I believe there are open bugs in PreloadJS and SoundJS which prevent audio loading from reporting progress. I have logged issues for both libraries:
https://github.com/CreateJS/PreloadJS/issues/99
https://github.com/CreateJS/SoundJS/issues/119
Some additional info:
Although PreloadJS will try and favor XHR-based loading for filetypes it controls, audio loading works a little differently.
SoundJS actually injects the functionality into PreloadJS to handle loading, and will not respect the useXHR parameter, instead relying on the browser capabilities (and SoundJS plugins) to load and play audio. Unfortunately, web audio requires an array buffer (loaded with XHR), whereas HTML audio requires HTML tags, so the playback capabilities dictate how audio files load.
By default, SoundJS will default to load/use the following plugins in order:
WebAudio (therefore XHR)
HTML (therefore tag-loading)
This should favor XHR-loading and webaudio for most browsers (IE is the standout that will almost always require HTML loading). You can force plugin order by registering the plugins manually before you begin playback/
Thanks for the surfacing this!
#Lanny is correct, there are open bugs for this issue. Currently SoundJS is setup in a way that always uses Tag loading with PreloadJS, even when using WebAudio which loads via xhr. The result is that regardless of plugin, there is currently no way to get progress events.
The good news is that we are currently in the process of revising how loading works between PreloadJS and SoundJS and this issue should be resolved.
I'm trying to automate a work flow where we have to log in to a website, navigate, get redirected several times and finally have to upload a file into a reporting system.
After failing with phantomjs/casparjs (where we also do not really get visual feedback) I was thinking about using node-webkit.
So basically, what I am trying to do is writing a "controller" that is opening another webpage in an iframe and then manipulating the fields, hitting buttons, ...
Is this something that can be done? If yes, I am struggeling to get a handle on the fields to fill them...
Or is this a classic "wrong tool" approach and we shouldn't be doing that?
Something along the line of
var new_win = gui.Window.get(
window.open('https://remote/login/site/')
);
gui.Window.get(new_win).on('loaded', function () {
//all of this doesn't really work but might help you to understand what I try to do
//window.console.log(new_win.window.document.getElementById("user"));
//window.eval(new_win, "code_to_fill_the_user_field");
//var userField = new_win.window.document.getElementById("user");
//console.log(userField);
});
Update: 2014-08-02:
I understand now that webkit is intended for creating desktop apps with HTML5 and not remote controlling websites, so we forget about this question.
I did solve the problem with phantomjs/casperjs now, BTW.
I understand now that node-webkit is intended for creating desktop apps with HTML5 and not remote controlling websites, so we forget about this question. I did manage to solve the problem with phantomjs/casperjs now.
I implemented webKit into a c# app i'm using and it works great except for one problem. Let me give a background of what i'm using it for.
This is a program i've been building that will run multiple tests on multiple customer accounts, it loads each test into it's own dynamically created browser controls and such. However the issue i'm having is that some of the tests use Iframes which pass information in between the parent and child pages. These Iframes run a few ajax commands that show that they are loading, however it doesn't finish executing whatever scripts it needs to run. Is there something i need to change in the way I've added WebKit to my application or anyone else that has had any similar problems? These tests work in chrome and safari as well so i know webkit can do it, but for some reason it's not doing it for me.
Thanks for the help!
I'm trying to get screenshots from a web page with multiple browsers. Just experimenting with Selenium RC, I wrote code like this:
var sel = new DefaultSelenium(server, 4444, target, url);
sel.Start();
sel.Open(url);
sel.WaitForPageToLoad("30000");
var imageString = sel.CaptureScreenshotToString();
This basically works, but in most cases the screenshot is of a blank browser window, because the page is not yet ready for display. It kind of works if I add a sleep just after the WaitForPageToLoad, but that slows down the fast browsers and/or may be to short for the slower browsers (or under load).
A typical solution for this seems to be to wait for the presence of a certain element. However, this is meant as a simple generic solution to get a screenshot of a local web page with as many browsers as possible (to test the layout) and I don't want to have to enter certain element names or whatever. It's a simple tool where you just enter the Selenium Server URL and the URL you want to test, and get the screenshots back.
Any advice?
I use Selenium-RC to capture screenshots of remote pages where the waiting time is variant. In such cases, checking the title of the page and using time.sleep(n seconds) usually does it for me.
May be you can make use of Browser status bar to verify whether that page is loaded fully or not. I think this is the best solution.