If I open http://chatphrase.com/test/ in two browser tabs (either on the same machine or different machines - I'm testing using Google Chrome for both) and enter the same word in both, the two tabs seem to perform the ICE / STUN / SDP handshaking successfully and connect to each other, but the remote video element, despite having the remote stream attached, remains blank (and hovering over the element causes it to shrink to 2x2).
https://apprtc.appspot.com/ works fine. What is my site doing wrong?
This seems to be an exclusively client-based issue, but for what it's worth, my server-side code is at https://github.com/stuartpb/chatphrase/blob/master/app.js .
When the ICE candidates come in, they're not being added to the peer connection:
https://github.com/stuartpb/chatphrase/blob/99ce85330594026d70d5f9441cc06be6c4904b7e/static/chatphrase.js#L44
https://github.com/stuartpb/chatphrase/blob/99ce85330594026d70d5f9441cc06be6c4904b7e/static/chatphrase.js#L76
https://github.com/stuartpb/chatphrase/blob/99ce85330594026d70d5f9441cc06be6c4904b7e/static/chatphrase.js#L168
addIce(resbody.ice); should be addIce(peercon,resbody.ice);.
(I wrote this function somewhere around 4AM, right around the time that you start forgetting the signatures of functions you just wrote.)
Related
Is it possible to capture desktop screen sharing through webRTC.. As we know that it just captures the screen on the browser tab but is it possible to capture the whole desktop screen like navigating through files on computer or opening and viewing files like pdf etc..
Currently, only "stateless" screen capturing is available in RTCWeb implementations (both chrome & firefox). E.g.
Install chrome extension and then try this demo
Above demo will simply capture screen of "any" opened application's screen. Though, such screen capturing API fails to capture screens of full-screen game applications.
More information available here:
https://developer.chrome.com/extensions/desktopCapture (HTTPs+getUserMedia+postMessage)
Regarding remote desktop sharing from a web-browser, it has a pile
more security risks associated with it compared to screen sharing. The
UI/security aspects will be tough to deal with, and the feature will
be very susceptible to social engineering -- phone call: "this is
Google/Dell/Computer-Management; we've detected your machine has a
virus on it; could you browse to and we'll assist you in
removing it" -- etc. Ref
Yes, it's possible. At least using Chrome. There are several ways of doing it, but the simplest one is:
Add this constaint when you invoke getUserMedia:
constraints.video.mandatory.chromeMediaSource = 'screen'
When starting chrome, use this argument (chrome version > 35):
--enable-usermedia-screen-capturing
You can find an example of sharing screen and recording the shared screen at a remote server repository here:
https://github.com/Kurento/kurento-tutorial-js/blob/develop/kurento-recorder-screen/static/index.js
If you try to execute that example, play close attention to the security restrictions. All signalling needs to travel using TLS. Using raw HTTP will produce chrome to refuse sharing screen.
Yes it is. I recently worked on WebRTC and was able to stream desktop easily. Following links helped me implement my requirements :
Firefox Extension : http://mozilla.github.io/webrtc-landing/
Dont forget to add your *.github.io to about:config -> getUserMedia screensharing allowed domains
Google Chrome extension : https://developer.chrome.com/extensions/samples#search
Open this in Run : Chrome.exe --enable-usermedia-screen-capturing
Other Reference : https://github.com/muaz-khan/WebRTC-Experiment
My WebRTC app works fine when I connect two of the same browsers, but when I try a combination neither respond to each others signaling messages. Something probably worth mentioning is that I have not implemented TURN, however I don't see why that should make a difference so I'm not going to change that unless I'm fairly certain it will.
I don't have much of a clue where the error lies, so I will just add code on request for the sake of readability.
Make sure you enable DTLS-SRTP (Firefox only supports DTLS-SRTP) by passing the following to the PeerConnection constructor:
{ 'optional': [{'DtlsSrtpKeyAgreement': 'true'}]}
See this page for more details.
You have not really described what goes wrong with the signaling. No error messages and so on.
But based on the fact that you only see the fault when using two different web browsers I would recommend using Adapter.js that have been somewhat promoted from webRTC.
Link to webRTC demo that shows on the interoperability using Adapter.js(page also contains a link to Adapter.js):http://www.webrtc.org/demo
Direct link to
adapter.js
Try to turn off your firewalls to check if it fixes the problem.
In my case (Windown 7), default windows firewall didn't allow UDP for Private Inbound Connection Setting and Firefox + Chrome p2p connection just didn't work.
Hope it helps.
When I browse the web with IE10 in win8's Metro part there is no problem but when I try to view page that is located on server in my local network(the same subnet) it displays this message:
This page can't be displayed
•Make sure the web address http://192.168.1.100 is correct.
•Look for the page with your search engine.
•Refresh the page in a few minutes.
If following these suggestions didn't work, resetting your connection might help.
Reset connection [<-a button here]
Get more help with connection problems
Now the funny part is that there is an option in metro version of ie10 to open page on desktop (in regular IE10) and than it works with no problem.
I can't find or think of any security setting that would restrict browsing websites inside your own local network.
(this is Windows 8 32Bit Release Preview build 8400)
Any ideas?
This is related to EPM (Enhanced Protected Mode) in IE10. It's hard to summarize in an answer here, but Eric Lawrence (a PM on the IE team) has an excellent post detailing everything about EPM:
http://blogs.msdn.com/b/ieinternals/archive/2012/03/23/understanding-ie10-enhanced-protected-mode-network-security-addons-cookies-metro-desktop.aspx
In particular, read the "Loopback-blocked" and "Private Network resources" sections.
In your case, you might try one of these approaches:
Try aliasing the dotted hostname (http://192.168.1.100) via a custom DNS entry (e.g. http://myservice)
Change the Trusted Zones settings
See if your network connection was established as sharing or non-sharing, which would trigger private vs. public mode.
Again, see Eric's post for the details of each of these.
I'm not really sure how to word this exactly, so hopefully someone can make sense of it. I've been working on an iPad app that syncs files from a server to your iPad and lets you build presentations with the various files. The corporation I'm working with on this app has a wireless network that requires you to re-authenticate every hour. So every hour instead of getting the expected JSON api response, any HTTP request pulls down the page needed to reauth with the wireless network. I was wondering is there is a specific HTTP response code related to getting sent that page or a "best-practice" way of testing for that page as opposed to JSON.
Granted I could just test to see if the response is HTML, but that doesn't account for other redirect responses that I haven't found yet. I could just test part of the HTML to see if it matches a predetermined portion of the html, but I'm an outside contractor. I can't guarantee they won't change the markup or verbiage of the page after I've made my deliverable.
So does anyone out in the ether know a "best practices" methodology for testing if the app needs to reauth before syncing?
I noticed that on Mac OS X and maybe even iOS, when you connect to a new Wi-Fi network, it will try to contact www.apple.com. This is being done to check if the internet connectivity is available. If it's not available, the Captive Network Assistant will pop-up, showing you the authentication page, or sometimes when I'm in Starbucks, an advertisement.
Following your question, since Apple themselves is doing it this way, I think you could check for HTTP response code, look for something in the HTML markup (slightly discouraged though) or trying to connect to a known server (Reachability).
For a corporation practicing well-documented projects, I am quite sure they won't be changing things without making sure that your app, once deployed will continue to work.
I can't seem to get Orbited working with my Twisted app. I have a page, served by Twisted (say localhost:8000/page) which includes Orbited.js from the orbited server (localhost:8001/static/Orbited.js). I then have a TCP chat server example running on port 7777. I try to use Orbited.TCPSocket to connect to the chat server:
conn=new Orbited.TCPSocket();
conn.open("localhost", 7777);
conn.send("test\r\n"); //error: bad readyState
It works fine when Orbited is serving the page, but not when twisted serves it from a different port. My orbited.cfg looks like this:
[listen]
http://:8001
[access]
* -> localhost:7777
And before (which worked) I had this in it as well:
[static]
test=index.html
Where index.html was another page grabbing localhost:8001/static/Orbited.js, and was accessed from localhost:8001/test.
How do I need to change my config file to work with requests from my twisted site on another port?
Update
I tried changing Orbited.settings.port to 8001 before trying to open the connection, but I got an error: "unsafe javascript attempt to access frame with url http://localhost:8000/page from frame with url http://localhost:8001/static/xsdrBridge.html#1. Domains, protocols and ports must match."
Hmm, also, I just looked at the orbited wiki, and apparently, setting Orbited.settings.port is exactly what I'm supposed to do. but I'm getting horrible errors
You can call send() only after the connection is in opened state.
Put a handler for .onopen() and do a .send() from there.
I have used Orbited in the past. It works in general but there are several quirks to get it set up and running smoothly. The project itself seems to be in a state of flux (it seems to be moving to node.js). Both of these points lead me to suggest that - if you can avoid it - not to use Orbited.
Are there alternatives that are cleaner? I would say, yes. You can pretty much emulate Orbited with Websockets on stock Twisted. This will clearly work for newer browsers. What about older ones? Well, there are open-source projects that wrap websockets and fall back to flash as a transport for older browsers. The setup works quite well, and actually feels cleaner than using a solution like orbited.
If you check out http://github.com/rlotun/txWebSocket you'll find the current state of Twisted's websocket implementation, as well as an example of how to fall back to flash on older browsers. Hopefully this will be useful enough for you to serve as a drop in replacement to Orbited.