I have the following situation:
Page A: /something/new
Which posts back to: /something/create
Which redirects to Page B: /something/edit
So far it all works. Now, /something/edit is a page that lets you do a bunch of things through AJAX, so it starts up empty, and as you use it it gets "fuller", so to speak. If you reload at any time, you get everything back, rendered by the server.
However, if after being redirected, and making modifications to the page, you hit Back and then Forward again, the browser (Chrome at least) doesn't hit the server again (not even an Etag check that might result in a 304, nothing), it just loads Page B from cache, which shows up empty, and can be quite confusing...
When first rendering Page B, the server responds with the following headers:
Cache-Control:must-revalidate, private, max-age=0
Connection:Keep-Alive
Content-Length:18577
Content-Type:text/html; charset=utf-8
Date:Thu, 02 Aug 2012 20:19:59 GMT
Server:WEBrick/1.3.1 (Ruby/1.9.3/2012-04-20)
Set-Cookie: (redacted)
X-Miniprofiler-Ids:["ma2x1rjc0kgrijiug5dj","nnmovj2wz1lux85jwhd3"]
X-Request-Id:2dd3fa62799beadc1b39b8db1aa5f45f
X-Runtime:0.245014
X-Ua-Compatible:IE=Edge
I don't see an Etag, or anything similar that could be bothering. Also, if I'm interpreting "Cache-control" correctly (i'm not very experienced with it, though), it seems to be saying to not cache, ever...
Is there any way to avoid this behaviour, and have the browser hit the server again on Back/Forward?
Thanks!
Daniel
I would investigate the answers posted here.
Is there a cross-browser onload event when clicking the back button?
I just changed this answer because some of the utilities I was referring to are pretty old and not maintained, which is not a useful answer.
Related
I am testing a login functionality on a 3rd party website. I have this url example.com/login . When I copy and paste this into the browser (chrome), page sometimes load, but sometime does not (empty blank white page).
The problem is that I have to run a script on this page to click one of the elements (all the elements are embedded inside #shadow-root). If the page loads, no problem, script is evaluated successfully. But page sometimes does not load and it returns a 404 in response to an XHR request, and as a result, my * eval(scrip("script") step returns "js eval failed...".
So I found the solution to refresh the page, and to do that, I am considering to capture the xhr request response. If the status code is 404, then refresh the page. If not, continue with the following steps.
Now, I think this may work, but I do not know how to implement karate's Intercepting HTTP Requests. And firstly, is that something doable?
I have looked into documentation here, but could not understand the examples.
https://github.com/karatelabs/karate/tree/master/karate-netty
Meanwhile, if there is another way of refreshing the page conditionally, I will be more than happy to hear about it. Thanks anyone in advance.
First, using JavaScript you should be able to handle shadow roots: https://stackoverflow.com/a/60618233/143475
And the above answer links to advanced examples of executing JS in the context of the current page. I suggest you do some research into that, try to take the help of someone who knows JS, the DOM and HTML well - and you should be find a way to know if the XHR has been made successfully or not - for e.g. based on whether some element on the page has changed etc.
Finally here is how you can do interception: https://stackoverflow.com/a/61372471/143475
I have a simple test web site with 2 different pages, Page1 and Page2. Pages are generated by a web server.
All responses contain these headers:
Cache-control: no-cache, no-store, must-revalidate
Pragma: no-cache
Page1 contains a link to Page2 and a TextArea element, enclosed in a FORM tag. Page2 is a blank page (it doesn't contain anything useful).
Clicking on the Page2 link (inside Page1) will submit the form (and TextArea value) to the server and switch to Page2.
Each time Page1 is requested from the server, a new line is added to the TextArea. This happens at server-side (no DOM manipulation via JavaScript, at browser side).
So, when application starts, Page1's TextArea contains:
"Line 1"
If I click on the link to Page2 and then hit the back button, Page1's TextArea now contains 2 lines, as expected:
"Line 1"
"Line 2"
This shows 2 things:
BFCache (Back-Forward cache) is not being used (because each time the back button is hit, a new request is sent to the web server). It can also be confirmed by the fact that pageshow's event parameter persisted property is false.
The browser is updating the page correctly (because new content added to the page, at server-side, is being shown at browser side, as expected)
However, if I clear the text area, click on the same link to Page2 and then hit the back button, the TextArea will still show an empty content. I would expect it to show exactly one line of text (inserted by the server). This happens in all tested browsers, including WebKit, FireFox and even Internet Explorer.
Using Network tab from Developer Tools I can see that the server responds with the correct content and the preview sub-tab (in dev tools/network) shows exactly that.
So, in the second scenario, the browser is definitely retrieving the updated content from server, but refuses to update the page with that content. Seems to me that whenever the page is submitted, the browser will use the submitted page when the user hits the back button, regardless of content updated at server-side.
This issue is very similar to another here: How to prevent content being displayed from Back-Forward cache in Firefox?
with same symptoms but in that case there is no form submit and change in response headers would fix it.
My questions:
a) Why browsers retrieve an updated content from the server but don't use it to update the page in that particular scenario? Is it a documented behavior?
b) Is there a way to change this behavior? I'm particularly thinking about a way to force the browser to update the page with the response from the server (the same way it occurs when no submitting is involved).
BTW, submitting the form in an AJAX request is not a solution but merely a workaround for this issue, having in mind that a full post back may be needed in some scenarios.
I have a vendor web application that sits within a frame being viewed with IE 11 on both Wondow7 and Windows 10. Emulation mode is set as default (Edge). The application functions properly most of the time. However, the application will randomly displays the error page "This content cannot be displayed in a frame". The headers being set are:
X-Frame-Options: SAMEORIGIN
X-Powered-By: Servlet/3.1
X-XXS-Protection: 1; mode-block
All content is coming from the same domain.
Everything I found on this topic so far relates to an issue occurring when the page is initially loaded not at some random point in time after page load. It also seems to occur not when the user is on the page but when they shift focus to another page or browser tab then return.
Any Ideas?
You could try to change the X-XSS-Protection value to 0. From this answer, we can see that:
The token mode=block will prevent browser (IE8+ and Webkit browsers) to render pages (instead of sanitizing) if a potential XSS reflection (= non-persistent) attack is detected.
So this might be the reason why the content can't be displayed sometimes.
Besides, if you're trying to open HTTPS page from non SSL site, you can also get this kind of error. You could refer to this thread.
I have a PayPal button which upon checkout completion redirects the user back to the homepage but in the background opens another php page which does a couple of things including updating some databases etc. However in it it I'd also like to perform a rest API request.
http://your-server/rest/createUser?username=testuser&password=testpassword
Here is the URL I need to fire and I'm not too bothered about reading the contents of this to be honest, it respond an XML file saying status OK or error messages but I don't need this... All I want to do is access the URL. This works from a browser, but when I try to use
$apicall = file_get_contents("http://your-server/rest/createUser?username=testuser&password=testpassword");
It doesn't work... Any thought ?
For file_get_contents() to work correctly any special characters have to be encoded.
$url = 'http://your-server....';
$encodedUrl = urlencode($url);
$apicall = file_get_contents($encodedUrl);
The other thing to check is that allow_url_fopen php.ini setting is set to true. (It is by default).
Here is the brief detail of issue.
I have Page1 where I have put LinkButton. The LinkButton Has property PostBackUrl pointing to Page2.
When user is redirected to page2, I am using Page Load method to access controls from previous page & get the needed value. To make clear, I am using this approach becuase I cant use querystring.
Page 2 has 2 web parts on it. The web parts use data received in Page Load event from page1 and renders data.
This works perfect on first page load. When user clicks on a URL in page which posts back, the web parts gets lost.
Note that if I come directly to page2 without going at page1, then web parts are retained in the page and they are not lost.
Can anyone give me the clue of issue cause?
Thanks in advance.
Do you have any debugging enabled? you are most likely looking for values on page load that don't exist and might be getting exceptions that aren't handled properly.
I am not sure why but somehow code was throwing an exception when I tried to access the property Page.PreviousPage. Though I had made sure to check null on each step. Even code was never get hit when web part was lost. So it is still a mystery for me.
Just in case someone comes across this issue my workaround may help. I used Post back to page2 using post method. I accessed the variables using Page.Form[] variables. This way my issue of getting web parts lost got resolved.