This question has been asked before but 1) the user never accepted an answer 2) none of them stand out as better than the others (votes-wise) and 3) the asker seems to have forgotten about it. So I'm going to ask it again so I can get to an accepted answer. And some of the users in the thread have said that some of the solutions didn't work. Sorry for cluttering up the place, but I promise to get to the bottom of this.
I ran into this problem the other day when I was looking at my Silverlight app in Firefox. I made a change to the location of an image and it didn't move. I assumed I did it wrong, but then I looked at IE7 and the image was in the right place. Turns out Firefox was displaying a cached version of the file; the changes I made didn't show up.
This is a larger problem: if I change my app (let's say it's an urgent typo correction) how can I force the end user to see the most current version of my Silverlight app? Is isolated storage (Heuer's blog) really the only way to force an update from the server side? Clearing the Firefox cache is not going to work for a push update; I need the update to propagate without the end user doing anything.
Update: Dino Esposito has some ideas about controlling this, specifically using the Expires property of the Response object. Haven't had a chance to try this yet.
Can you encode the version number or timestamp in the filename? That way, if the page changes, Firefox will notice that it points to a completely different resource and will reload it.
First, you need to build your application after every layout or code change. Silverlight is not HTML, it's code that runs locally.
Second, the actual solution to this:
page.xaml loads, kicks off an async on the web service that sits on the underlying page
webservice detects Firefox
if Firefox, tell the response that it expires immediately
[OperationContract]
public bool DetectFirefox()
{
if ((HttpContext.Current.Request.Browser.Browser == "Firefox") &&
((HttpContext.Current.Request.Browser.MajorVersion >= 2)))
{
HttpContext.Current.Response.Expires = -1;
}
return true;
}
The only caveat here is that you'll need to add that code before you want it to kick in. Otherwise FF3 will be running the old version of your code that doesn't have the response expiration.
Related
I have a create-react-app build being served by flask. It works fine in other browsers; however on Firefox, when I do a reload, the main.js file is stuck at the fetch method. After 3 minutes, it timesout then renders what it suppose to. I am looking for guidance on where to look from hereon. The production server does not log that the endpoint is receiving the fetch request; however, once the 3 minutes is up, the client side renders the updated information (as if it hit the server side). Everything looks like it should be working in the code I wrote, but I am not able scope out where I should be looking in debugging this issue.
Note: I am giving a broad description of the problem at the moment, but will disclose specifics such as codes, logs, and others items once I am able to get some guidance of the matter. As of now, there are no resources pointing out this specific issue, thus displaying code will simply be a "garbage" dump.
After working on the problem for quite some time, we found these two articles to be very helpful:
https://www.andreasreiterer.at/fix-whitescreen-static-react-app/
https://blog.miguelgrinberg.com/post/how-to-deploy-a-react--flask-project
After testing different sources, we first found that article one's suggestion of changing the homepage to the base url worked. After realizing that the nginx was getting confused on which static files to server, we proceeded to follow article 2's steps.
We did not have any issues on other browsers, but only on firefox. I hope this will help those that are experiencing the same Firefox issue.
First, I apologize that I don't fully understand what's going on here. I am new to AngularJS and I am building an app that I am running on localhost:3000. I make changes to my files, and I refresh the localhost:3000 page.
But after a few times of doing this, my changes do not get passed to the server.
Upon inspecting the developer tools, I see that it is using an older version of the file that I have overwritten with new code. No matter how many times I refresh the page, the changes do not get passed to the server.
If I come back after a few hours, it gets all the changes like it should. But only the first time, because every time after that I have the same problem.
I've tried restarting the http-server, I've tried closing and opening the editor, closing and opening the tab, restarting my computer, but none of these fixes the problem permanently.
If you need any more information, I'll edit it into this question. Thanks!
Try clearing your browser cache
Either
ctrl + f5
or
Right click on refresh button and choose the last option (hard reload)
(only in chrome, I guess)
That's almost certainly a caching issue.
You need to look at the network traffic when the page is downloading.
How you do that will depend on the browser you use, but try pressing F12 for a start
Check the RESPONSE headers for anything "cache"
Also check for a Status code of 304 - "Not Modified"
That might reveal to you instructions from the server to the browser to cache the file for a few hours, or that the browser is being told the copy it has is up to date.
Alternatively it might be server side caching, in which case I can't help you much.
One other solution is to change the file reference to include the date modified as a query string.
e.g.
instead of
src="/scripts/myscript.js"
use
src="/scripts/myscript.js?dt={filemodifieddateformyscripts.js}"
I've ran into an issue with Safari on iOS 5.1.1 and Angular where I am doing paginated results. After I decide to leave the page that is hosting the app, and then come back to the page by using the browser's history back button.
The application is not initialized so as I scroll again, the pagination functionality is not triggered.
Since 5.1.1 doesn't provide a web inspector, I had to rely on console.log to print events that I want to see that it's executing. On initial load, I see everything ok, but when I am back through the back button nothing fires.
My usual ways of fixing this are not working. Here are the things I've tried:
Used manual bootstrapping of Angular app
Made the page not support caching and expire the page as soon as it's rendered.
Neither of these helped, which is why I am here asking for some suggestions. Unfortunately it's not a very easy thing to replicate through a Fiddle or a hosted HTML file so I am unable to provide one. Hoping that some of you may have experienced this before.
Thanks in advance for any help.
Answer is quite late and I hope you found a solution already but it might be useful for others.
Unfortunately I won't be able to answer why the app is not initialized (due to the lack of code) but I can give you some tools/tips to ease the debugging (assuming you have the iOS simulator).
Debuging on mobile safari can be done either :
with weinre and get the features of the web concole,
or by enabling remote debugging in Mobile Safari itself, see here
Yea, same problem here with iOS 7. It must send the DOM structure over before running any javascript. Kinda pointless...
Since the problem is with mobile safari and since it supports popstate event, you can use this event to detect when the user is back in combination with $location.path(), for example:
$(window).bind("popstate", function () {
alert(location.pathname + $location.path());
});
Over Christmas I implemented some code to open a channel to my App Engine application using the channel API and it was working fine.
I recently returned to the code and find that it is broken, and the problem seems to lie in talkgadget that the client code is trying to load into a frame.
I took a snapshot of the code returned by /_ah/channel/jsapi on December 21 and I am comparing it to today's code.
When the older version tried to get talkgadget.google.com/talkgadget/d?token=[my-token] it was hardcoded to use http. Sensibly enough, the code was later changed to check the page's protocol in document.location and adapt to cases where https is used. The problem now is that when running in a chrome extension it grabs chrome-extension: as the protocol and naturally the channel fails to open in the extension.
Setting the base of the background page to my appspot URL does not change document.location, although I had some vague notion that it should. (Not sure what the standards say.)
Is there a workaround for this? One thought was to create my own iframe loaded from my site, but then it will nest the actual channel iframe, complicating things.
I wonder if Moishe or someone else on the App Engine team could comment on this. It seems like the this is a bug and the code should take into chrome extensions either by testing for the protocol and searching for a base tag as an alternative, or provide some way to configure this through the javascript object.
Yep, this is a bug. There's a fix submitted, and it's in the process of rolling out to production; it'll probably be out in about a week.
In the meantime, your idea of an iframe embedded in the extension should work.
I'm assuming the issue I'm having is related to caching. Code changes I make are not getting picked up when I debug. Most times I get served a previous version of the app. How do I prevent this from happening?
Ctrl+F5 is an easy way to refresh a page and clear the cache of that page at the same time - it may help :)
Try to add to the page that hosts Silverlight application on Page_Load:
Response.Cache.SetExpires(DateTime.Now.AddSeconds(-100));
Response.Cache.SetCacheability(HttpCacheability.NoCache);
Append a "version" querystring to your XAP Url, something like:
http://localhost:1234/ClientBin/my_silverlight_app.xap?v=1.0.287.5361
This will trick the browser (and many web servers) to think that this is a different file. And when the cache problem appears again, increase the number.
If you then want to employ proper caching, do it on the server-side with OutputCache directives.
As far as I see, this seems to be a problem with Firefox - when I used IE8, this didn't happen to me (I realize this may open its own can of worms, but at least for debugging and testing Silverlight, IE is much better)
I have not had any issues with Silverlight assemblies getting cached - you might want to try debugging the HTTP requests that go back and forth, to see if maybe your server is instead returning incorrect information to the browser (e.g. a "not modified" response).
For general no-cache behavior, the only reliable method I have found is to turn off caching in the browser.
For IE, this has been the only reliable option - otherwise, even if proper no-cache headers are sent, certain things are still cached (specifically, dynamically loaded resources which are accessed via Javascript XmlHttpRequest). I have not specifically had issues with Silverlight getting cached when it should not, though - IE has always loaded the latest updates even if cache is enabled.
Firefox has been much more problematic - even when disabling cache, it still sometimes caches XmlHttpRequest-loaded resources. Manually hitting Refresh a few times has been the only solution in such a case. Once again, I have had no issues with Silverlight assembles, even if cache is turned on.
In Firefox, I use the 'web developer' plugin and simply select to 'disable cache'. Works fine.
Firefox 3.5 under Tools has the option for Private Browsing. Click that to disable caching.
Here is how I have done it for flex/flash and silverlight and it works.
Code Behind ASPX or CSHTML
string slUrl = "/ClientBin/MySilverlight.xap";
string filePath = Server.MapPath(slUrl);
FileInfo info = new FileInfo(filePath);
// this will force browser to
// re download file if file was
// updated
slUrl += "?t=" + info.FileWriteTime.Ticks;
ASPX or CSHTML
<embed ....
src="<%= slUrl %>"
..
/>
Trick is you have to change url by adding something after ? and make a new arbitrary random query string or use file write time, and for browser, something?t=1 and something?t=2 are two urls and it will not pickup cache if t changes.
Instead of write time, you can also choose any standard config value or you can even simply hardcode your ASPX or HTML and append something after ? that will force browsers to download silverlight xap file again.
<embed ....
src="/ClientBin/MySilverlight.xap?something-different-each-time"
...
/>