Why does Jitsi Meet totally peak out my CPU, even with the camera turned off, and nobody else in the room with me? They must be running some bloated ass Javascript
I also notice that it take a few minute before my CPU (dual core 1.6Ghz) goes back to normal after I close the #JitsiMeet tab. Hmmmm ...
@strypey hey I've been wanting to test Jitsi. I often use ZOOM. It's proprietary but has great a feature set. I'd 100% prefer to drive eyes to something like Jitsi. Your experience sounds like I might need to use some caution.
@david_ross be aware that my computer was new when the Ark was being built ...
https://www.coactivate.org/projects/disintermedia/bishop
@david_ross , but I tried logging in with #OpenBox, instead of #GNOME #Fallback, OB uses a bunch less CPU while idlng, and Jitsi Meet still pushed the CPU to 99% without doing anything but idlng with no cam. I also tried the #FramaSoft instance (#FramaTalk) and it was just the same
@david_ross I'm excited about testing #NextCloudTalk, the new #WebRTC app for the NC platform. I believe it's a rewrite of #Spreed, so I'm hoping that means it's a bit more resource efficient than #JitsiMeet
@strypey I brought this up on a Mozilla vols call today. Thanks for the update!
@david_ross @strypey Please kill #WebRTC with fire until it is a crispy charred hunk of Zuul meat.
Thanks.
@Shufei @david_ross I don't know, #WebRTC is pretty useful. #WebTorrent is built on it, for example
@david_ross it could be #XMPP partisans grumpy about #WebRTC eating their lunch?
Aside from being a hudge attack surface, it tend to leak a lot of informations about both your local network and your video/audio devices. This together with webgl can give trackers a pretty accurate fingerprint of your computer.
Luckily, ublock origin can prevent most WebRTC leaks. It's an opt-in option though.
You can test that here: https://browserleaks.com/webrtc
@Ninjatrappeur @david_ross hmm. Is it possible that the same thing that allows #uBlockOrigin to stop #WebRTC leaking private data could also stop sevice based on it from working properly? Especially in combination with #NoScript?
@Ninjatrappeur @strypey whilst not at all ideal this is my preferred method on anything I do not trust. New user profiles are the quickest use I've found.
Seeking more info on some of those issues you spoke of. Complexity is not something of concern for me. The fingerprinting and attack surface are of course.
I am a little concerned of the blurring of this from the 2016+ drafts - and seek up-to-date insight on how IETF & browsers in general have moved since. Will update you.
