When I try to debug my gwt app that's inside an iframe (note that gwt codserv and webserv are all local) I'm having permissions issues. The following errors are thrown:
on Chrome console: "Refused to display document because display
forbidden by X-Frame-Options.";
on Chrome webpage window: "Plugin
failed to connect to Development Mode server at 127.0.0.1:9997 Follow the underlying troubleshooting instructions"
This ONLY happens in Google Chrome, firefox and IE are ok.
I've already searched for simillar problems but all of them are on crossdomain context, and mine is local. Also, tried the proposed solutions but all without success.
Here is an example of my webpage containing my iFrame. I can't debug MyGWT app.
<!doctype html>
<html>
<head>
<meta http-equiv="content-type" content="text/html; charset=UTF-8">
</head>
<body>
<iframe src="MyGWT.html?gwt.codesvr=127.0.0.1:9997"></iframe>
</body>
</html>
The page that's being blocked is probably TroubleshootingOOPHM
FYI, this has been fixed in GWT 2.5, which now links to this page rather than trying to display it in an iframe (moreover underneath the error glasspane, which made it hardly readable anyway before Google changed their server settings)
Related
UPDATE:
It appears the problem is triggered when port 443 is explicitly specified. If I leave it out, no errors. For example:
<!DOCTYPE html>
<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
<title>Loader Test</title>
<script type="text/javascript" src="https://127.0.0.1/myapp/_vVERSION/dojo/dojo-release-1.16.3/dojo/dojo.js"></script>
<script type="text/javascript">
dojo.registerModulePath("myapp", "https://127.0.0.1/myapp/_vVERSION/myapp");
// The following should create the myapp object/namespace -->
dojo.require("myapp.bootstrap");
myapp.webContext = "https://127.0.0.1/myapp/_vVERSION/";
</script>
</head>
<body>
HELLO WORLD!
</body>
</html>
So, for whatever reason, if :443 is added to the URLs, an error occurs trying to set myapp.webContext.
ORIGINAL POST BELOW:
I am on a project that has a legacy code base that uses the non-AMD Dojo Toolkit method for loading modules. The application runs in Tomcat, and I am testing out running it as an app server behind an Apache httpd reverse proxy (gateway) with SSL and authentication enabled.
When I set the gateway port to a non-standard value (e.g. 4443), everything appears to load and work fine. However, if the gateway port is 443, I get javascript errors that indicate that dojo.require() statements are happening async vs sync, so errors occur in code that tries to reference objects from the required modules.
I have condensed things down to simple test page that recreats the problem (which occurs in Chrome and Firefox):
<!DOCTYPE html>
<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
<title>Loader Test</title>
<script type="text/javascript" src="https://127.0.0.1:4443/myapp/_vVERSION/dojo/dojo-release-1.16.3/dojo/dojo.js"></script>
<script type="text/javascript">
dojo.registerModulePath("myapp", "https://127.0.0.1:4443/myapp/_vVERSION/myapp");
// The following should create the myapp object/namespace -->
dojo.require("myapp.bootstrap");
myapp.webContext = "https://127.0.0.1:4443/myapp/_vVERSION/";
</script>
</head>
<body>
HELLO WORLD!
</body>
</html>
The above works with no problems. However, if I change Apache httpd configuration to use the standard https port of 443, and change the 4443 to 443 in the above, I get an error when trying to set myapp.webContext. When using the Network profiler in the browser debugger, it shows the request to the myapp.bootstrap as Pending, not loaded yet. If I configure the debugger to not pause on exceptions, the Network tab shows the module eventually gets loaded.
The use of absolute pathnames in the application is to ensure proper loading of resources and link resolution. This was required a long time ago due to problems in IE, and a configuration setting is supported to indicate the base URL for all hrefs (for cases when running behind a gateway).
As a test, I redid the sample to use all relative pathnames, and the page appears to load with no errors when using port 443. Unfortunately, the real application is quite large, so changing it to always use relative pathnames is not a viable solution at this time.
My gut is telling me the browsers are behaving differently with XHR requests that involve standard ports (e.g. 443) and use absolute URLs. Can anyone confirm this? And if so, why they behave that way?
I am using the following Content Security Policy in Report-Only mode:
Content-Security-Policy-Report-Only "default-src 'self'; report-uri /log_violations"
When I go to a URL on the server with an HTML page containing the following:
<!DOCTYPE html>
<head>
<title>
Test document
</title>
</head>
<html>
<body>
Hello
</body>
</html>
I am seeing the following error message in Firefox 57.0 on line 1:
Content Security Policy: The page’s settings observed the loading of a
resource at self (“default-src http://www3.thestar.com”). A CSP report
is being sent. Source: ;!function(){var t=0,e=function(t,e){ret...
In other browsers, like Edge, Chrome, I haven't seen these errors.
Any thoughts on whether this is a Firefox quirk or something I've set up incorrectly? I'm stumped as to why the policy is rejecting line 1 of every page.
You probably have an extension installed and enabled that is injecting content into the page. Try opening the page in a new Firefox profile without any extensions to see if you see the CSP violation reported.
Firefox is planning to exclude content injected by extensions in CSP checks starting Firefox 58 so this should reduce the noise from extensions doing stuff on a page.
From the Mozilla blog,
Starting with Firefox 58, the CSP of a web page does not apply to
content inserted by an extension. This allows, for example, the
extension to load its own resources into a page.
I am working locally on a PHP site. I am on a Mac OS X Yoesmite and have setup Apache. But sometimes when i refresh the page it doesn't load, so I have to refresh the page again and again until it randomly loads my site again. When it doesn't load and I view the source it looks like this...
<!DOCTYPE html>
<html>
<head>
<meta charset="utf-8">
<title></title>
</head>
<body>
<!-- ERR_DNS_FAIL -->
</body>
</html>
Anybody know why this is happening or how I might go about fixing it as it's very annoying when trying to build a site and you have to keep refreshing the page until it loads.
I know(pretty sure) what was causing this issue now. I switched browsers developing because I noticed it was only happening in Chrome, so I was using Firefox. But then when I installed "Zapyo" browser extension in Firefox(I already had it in Chrome) the same thing started happening. So I just disable this plugin when developing.
So nothing to do with Apache configuration.
I'm trying to embed a pde made with processing into a vanilla html page:
Here is my code:
<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
<script type="application/javascript" src="processing-1.4.1.min.js"></script>
</head>
<body>
<canvas width="700" height="821" data-processing-sources="jailmap2012capacity.pde"></canvas>
</body>
</html>
When the page is loaded, nothing appears.
I'm not getting any errors showing in the console for Firefox, but I am getting an error in Chrome that says, "Uncaught TypeError: Cannot call method 'addListener' of undefined" for client.js on line 6, but I don't think this is related to the problem.
The processing pde renders correctly within the Processing IDE.
Ideas? Suggestions? Thoughts? Thanks!
EDIT: Changed "source" to "src" in script link.
SOLUTION:
It appears that the default .pde produced in the Processing IDE does not include preloading tags necessary for use on the web.
For reference: http://processingjs.org/reference/preload/
Image and fonts will need to be preloaded on the very first line of the .pde, like so:
/* #pjs preload="path/image_1.gif, path/image_2.gif"; font="path/font_1.eof, path/font_2.eof"; */
javascript is not "application/javascript", it's "text/javascript", and if you're using , leave off the type. It's not necessary. On which note: add at the top. With that covered, if running that using localhost or online and that doesn't fix things (you cannot reliably run from file:/// unless you tell your browser to allow file XHR from file), check your console log, and report back what the errors are.
Once I asked this same question but now I can give a test case for it.
The problem resides in debugging in Eclipse PDT when the page sends multiple requests to the server's dynamic pages. Consider a web page as follow:
<!DOCTYPE html>
<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8"/>
<script src="link-to-jquery/jquery-1.7.2.min.js"></script>
<script>
$(document).ready(function() {
var list = $("ol");
for (var i=0; i<20; i++) {
list.append($('<li><img src="/img.php?'+i+'" /></li>'));
}
});
</script>
</head>
<body>
<ol>
</ol>
</body>
</html>
In above page, JQuery is used only to prevent browser from caching the images. And the the img.php reads like this:
<?php
readfile('some_image.jpg');
When I try to debug the first page in the Eclipse PDT, using Zend Debugger, only the first img.php request is made and others are dismissed. A sample output is shown in the attached image file. In the presented case, not loading an image file won't prevent you from debugging the rest of project. But once there's a javascript file which isn't loaded because of this problem, the rest of project won't work as it has to. Does anyone know how can I debug such a page?
Here are my specifications:
PHP Version 5.3.14
Zend Debugger v5.3
Eclipse for PHP Developers, Version: 3.0.2
Apache/2.2.22 (Ubuntu)
I found that this problem is specific to Zend Debugger and XDebug works smoothly.
+1 for XDebug, -1 for Zend Debugger