Before jumping in with an answer, please make sure you understand my scenario.
I have ajax calls that CREATE flashes.
I have other ajax calls that FETCH the flashes as JSON.
What is currently happening: I click a button which creates the flash. After which I run a ajax call that executes:
public function actionGetAllFlashesAsJSON() {
$flashMessages = Yii::app()->user->getFlashes(true);
$returnResult = array();
foreach ($flashMessages as $key => $value) {
$newItem = array();
$newItem['message'] = $value;
$newItem['kind'] = $key;
$returnResult[]= $newItem;
}
print json_encode($returnResult);
die();
}
My problem is, when I execute this function twice in a row, it still keeps returning the flashes. However, if I refresh the site, it shows the error, and then if I press refresh again, it's gone. My theory is that page refresh is causing some other kind of deletion of messages... but what? And how can I force the deletion of these messages after I receive the message in the above code?
More background info: I am using the flashes as ERROR messages, but i want them to appear at the top of my site AS THEY ARE CREATED. Flashes might get created via Ajax, so I have javascript running to check for new messages, and display them, but my problem is it shows the messages several times, because they are not getting deleted after calling getFlashes?
The flash messages are controlled by SESSION variables, which Yii destroys when the page is loaded (probably somewhere quite deep in the framework). You will have to manually destroy all the previous flash messages at the start of the ajax request
You can use: getFlashes() to get all the existing flash messages
For the other flash message methods have a look at the CWebUser docs here
Related
Question about how to send a jQuery callback with an onSuccess: refresh from a textInput when the user presses [Enter]. We use the [Enter] press to trigger a search callback.
Our GS Seaside app uses HAProxy. This means the onSuccess: script is handled by a different gem than the one that handles the callback. Because of this, users will sometimes get the refresh because the callback, which to them looks like a lost input (a browser F5 refresh shows the right state). If running single gem or in a VW / Pharo image this problem does not come up.
I can work around the problem by using...
async: false;
...but that prevents me from show any kind of waiting feedback (I normally use a busy gif).
So, the question is: in a multi-web server configuration, how can you code a callback to...
1 - show a busy gif
2 - trigger the search callback
3 - refresh the display when done
...all in that order.
Using a form submission callback is a problem because multiple text inputs can trigger the search, since the callback is 'set value + do search', by virtual of the default [Enter] press.
For the JS callback, I'm using...
self onKeyPress: (
(JSStream
on: '(window.event ? window.event.keyCode : event.which) == 13')
then: (canvas jQuery ajax callback: aBlock value: canvas jQuery this value))
It all works fine, except for the missing busy gif, due to the 'async: false'.
Any suggestions?
You can define a beforeSend and a complete handler to show and hide the loading indicator while the request is being processed. The global parameter set to false is meant to ignore your existing handlers to process request start and end (the mentioned spinner), and only use these defined in the particular instance of the JQAjax object.
((html jQuery ajax)
async: false;
global: false; "https://api.jquery.com/category/ajax/global-ajax-event-handlers/"
callback: aBlock value: canvas jQuery this value;
onBeforeSend: (html jQuery id: 'indicator') show;
onSuccess: ((html jQuery id: 'fieldId') load html: [:h | ]);
onComplete: (html jQuery id: 'indicator') hide;
That said, keep in mind that doing a synchronous AJAX call is discouraged since it will block the whole UI thread until the request is resolved.
So it's not completely clear how you manage the state in different worker images (gems, in this case) returning different things (probably because of having different sessions), so it's also not clear to me why doing an async XHR request will be served differently to doing it synchronously, I never experienced that.
From the small sample you mention, it can't be deduced what is the "refresh" part of your code. So maybe, providing more context will help us give you more accurate answers.
Fix ended up being trivial: needed to include 'event.preventDefault();' in the [Enter] key script. Seems obvious in hindsight.
if ((window.event ? window.event.keyCode : event.which) == 13) {
event.preventDefault();
};'
This problem is confirmed to be a narrow configuration symptom: GemStone with multiple gems. On every other configuration Seaside / javascript behaves as expected. I will follow this up as a vendor problem. Thanks to everyone that looked at it (this was also posted on other forums).
I have an MVC 4 Razor intranet app using Windows credentials where a new user must register (existing users are automatically redirected to the main user page). This generates an approval request email to the admin and returns a view with an Exit button simply saying your application has been submitted. The user then closes the app.
The email is in HTML and presents the user's entered data and has Approved or Declined selections. It then has a button to invoke code to insert the user into the Members table (if accepted) and return an email to the user with the decision.
Since some time may transpire between the app mailing the admin and the admin making the decision, the session will originally terminate. I need to have the email invoke a method to persist the new user, email the decision, and then simply exit. It will not involve the browser in any way, so no Views are involved. I could write this as a background console app, but that would involve duplicating a lot of code in the MVC app - with referential integrity issues on bug fixes or updates.
How can I write a method in the Controller that can be invoked by a link in the email as if from a browser that does the work and then exits without returning anything to a browser?
If I make the method an ActionResult method and return new EmptyResult() or return(null) I assume it will try to return an empty page to nowhere.
Can I alternatively construct a method in my Controller, where it has access to all of the support code, like?
public void EnrollMember(Member member, bool decision)
{
if (decision == true)
{
// insert new user into Members table
// generate accepted email
}
else
{
// generate declined email
}
}
and then just link to it the same way I link to Index(), /MyController/EnrollMember(...)?
There is no main() in the app since it is an ASP.NET MVC app, and I don't know how to terminate the app from within a Controller method without trying to return something, instead of from a View in the browser.
I'm a long-time programmer, but a .NET newbie so this probably has a simple answer.
The "view" is just the response. You don't necessarily have to return a view, but you must return a response. There's no way around that. If the link opens in a browser, then something will be displayed, regardless. Even if you were to just return an empty ContentResult, at least a blank page will be displayed in the browser.
To achieve something akin to what you're looking for, your best bet would be to return an HTML document with just a simple bit of JavaScript that will act to close the window:
<html>
<body>
<script>
window.close();
</script>
</body>
</html>
I've a test case where after creation of an item, notification is shown, but that notification disappear s within 2-3 seconds.
I want to identify the element for that notification, but when I try to inspect element for that in firebug, it's HTML snippet disappears very quickly since notification itself disappears. Hence I couldn't identify element for it and finding it very difficult to automate.
Can anyone suggest how to deal with such situation?
Since you say "I've a test case", I'm thinking you are testing an application of yours. The plainest way to get around your problem is to know your application. Even if a third-party library is providing the notification code, you can read the doc to see if you can increase the delay or you can read the source code to figure out how it creates the element and where.
If the above fails, then if you can get together a sequence of operations in Selenium that triggers a notification, you should be able to get a serialization of body quickly enough that you can then examine at your leisure. Using Selenium for Python, it would go:
print driver.execute_script("return document.body.outerHTML;")
I would run the code above with redirection to save the output of the print statement to a file that I'd then examine at my leisure. You could make it narrower if you wish by getting the outerHTML of a descendant of body. I like to have a good bit of context so that I know where exactly the element is being created. I've used libraries and configurations that create such notifications as children of body, and some that put them as children of other elements.
Open the previous page before the notification
Press "Ctrl+Shift+c", Navigate to "sources" tab
Do the manual step to generate the notification, which is going to hide in few seconds
Press "F8". Page load scripts will be paused
Then inspect the element as usual and fetch the xpath at your convenience
You can use the below JAVA code to wait for 20 seconds till any element with text 'your notification' appears in the page:
try{
WebDriverWait wait = new WebDriverWait(driver,20);
wait.until(ExpectedConditions.visibilityOfElementLocated(By.xpath("//*[contains(text(),'Your notification here')]")));
}catch(Throwable e){
System.err.println("Error while waiting for the notification to appear: "+ e.getMessage());
}
I'm trying to write an extension that will block access to (configurable) list of URLs if they are accessed more than N times per hour. From what I understand, I need to have a start script pass a "should I load this" message to a global HTML page (who can access the settings object to get the list of URLs), who will give a thumbs up/thumbs down message back to the start script to deny/allow loading.
That works out fine for me, but when I use the usual beforeLoad/canLoad handlers, I get messages for all the sub-items that need to be loaded (images/etc..), which screws up the #accesses/hour limit I'm trying to make.
Is there a way to synchronously pass messages back and forth between the two sandboxes so I can tell the global HTML page, "this is the URL in the window bar and the timestamp for when this request came in", so I can limit duplicate requests?
Thanks!
You could use a different message for the function that checks whether to allow the page to load, rather than using the same message as for your beforeLoad handler. For example, in the injected script (which must be a "start" script), put:
safari.self.tab.dispatchMessage('pageIsLoading');
And in the global script:
function handleMessage(event) {
if (event.name == 'pageIsLoading') {
if (event.target.url.indexOf('forbidden.site.com') > -1) {
console.log(event.timeStamp);
event.target.url = 'about:blank';
}
}
}
I want to be able to dispatch a message from the global page to the content script running inside an IFRAME in one of the tabs. I found that the Safari extension API may have a design oversight that does not allow this. (As opposed to the Chrome extension API, which has more comprehensive support for messaging between various components of an extension).
Basically, I have an injected.js that is injected to all frames and their IFRAMEs. In injected.js, I have the following for catching messages from the global page:
safari.self.addEventListener("message", function(msg) {
/* do something according to the msg */
} , false);
In my global page, I have the following code that broadcasts messages to the injected scripts:
var broadcast = function(message_name) {
var ws = safari.application.browserWindows;
for (var i = 0; i < ws.length; i++) {
var ts = ws[i].tabs;
for (var j = 0; j < ts.length; j++) {
ts[j].page.dispatchMessage(message_name);
}
}
};
What I found is that the messages only reach the injected.js injected into the top frames of the tabs. The messages never reach those that are injected in the child IFRAMEs.
It appears that the SafariWebPageProxy class only dispatches messages to the top frame in the tab, in spite of the fact that the injected.js script is injected into the child IFRAMEs, and that they all register for the "message" event on safari.self.
Do you know how I can get the messages to the injected script in a child IFRAME, and that the IFRAME is from a different domain vs the top frame?
It appears that the oversight in the API is as a combination of the following flaws:
(1) communications are basically broadcaster-subscriber based. There is no point-to-point callback-based communication channel like Chrome's chrome.extension.sendRequest API;
(2) this would have been fine if the broadcast goes to all child IFRAMEs. However, that is not the case. SafariWebPageProxy.dispatchMessage() only send the message the to injected.js in the top frame.
Any suggestion?
I ran to the same issue. Background can send a message to an IFRAME only in response to an IFRAME message.
To get around the limitation, I've used polling: each IFRAME is sending messages regularly to the background page to give it an option to send a message.
I know this is really old, but I recently had to do the same thing and the situation with Safari has not improved.
Although not ideal, for safari you need to rely on window.postMessage().
The top level injected script acts as a relaying agent between any and all iFrames on the page, taking in messages and passing them on to the background as needed. As complicated as that it sounds, it's quite likely the only way of making communication between a cross-origin iframe and the background page possible.