Open farao opened 4 years ago
To draft a possible solution:
<button>
s are implemented (those are the buttons that appear on the videos) and add one for the generated logo (using the function from 1. - maybe until now with a random seed or so).Maybe first it could be only a logo or only a color - then the probably not so easy step to apply a color to the logo would be out of the way
Encryption keys are not persistent to prevent fingerprinting.
Hm, ok, makes sense! I guess then we should probably use a hash of different information like the user agent. Or is it bad to allow users to recognise other people by this (e.g. if you chat 2 times with the same person/device) and we should just make it completely random?
Fingerprinting without the direct knowledge of the user and the ability to circumvent doesn't seem right to me. We could store a random value in local storage to keep the identity (gives user the ability to clear it and interacts well with browsers private mode). This mechanismwould probably need a method to prevent spoofing of that value (can't just send that value directly as it could just be copied).
I personally would prefer not to include such a non-obvious way to track the user and user a random non-persistent seed or a name that can be changed by the user.
Possible solutions are:
1. let a user choose a name (and probably generate one for those who don't choose one) 2. choose some other identifier - I propose to have a combination of a logo and a color.
I would also prefer option 1 for now, generate a random name. Could also be something more sophisticated, like a random avatar.
Also note, currently all peer colors are random for every participants, everyone sees different colors. We could implement a mechanism to map a specific colors to the peers. If we do so, this should be accompanied by a name or avatar, since relying only on colors is bad accessibility-wise.
avatar is what I mean by "logo". In any case I think colors might help a lot. I just though that choosing a name makes the whole interface more complex (and we need to find space to display the name), that's why I suggested avatar+color.
There are at least two different cases in which we would want to identify a person by some unique "identifier":
Possible solutions are:
Choosing 2 could also have a security aspect: If it's possible to get a fingerprint or the whole public key of every participant - we may choose a logo/color combination matching that and so we see the same symbols/colors every time we chat with the same person (on the same device).
I would imagine this to appear as one more "bubble button" on the videos (or the audio field when only audio).