- cross-posted to:
- hackernews@derp.foo
- cross-posted to:
- hackernews@derp.foo
cross-posted from: https://lemmy.world/post/3301227
Chrome will be experimenting with defaulting to https:// if the site supports it, even when an http:// link is used and will warn about downloads from insecure sources for “high-risk files” (example given is an exe). They’re also planning on enabling it by default for Incognito Mode and “sites that Chrome knows you typically access over HTTPS”.
You must log in or register to comment.
Don’t use chromium based browsers.
Is this just general advice? If so, I agree, but if it’s specific to this, what’s the problem you see with it?
It is general advice, but https should have been the default for a good while.
If we have to pick just one reason: WEI. As someone who’s been a professional software engineer for a decade and a half, this has the potential to mutate and ruin the internet at large in ways we’re only beginning to fully explore and understand.
Why just one? :p
Second reason: killing manifest v2 to kneecap ad-blockers.
Third reason: banning ‘trackmenot’ extension from the chrome store since its purpose is to muddy search stats (enhancing privacy, but in a way that messes up Google’s ad metrics and their history of your preferred search terms).
Keep grinding that ax.
I will. It’s sharp now. I will make it sharper.
I’ll keep using Brave because I like it. Thank you for your concern.
Chrome didn’t already default to https? Why?
It does if you just type in something like wikipedia.org . This most recent change they’re working on is so that a link on a page to:
http://wikipedia.org will get redirected to https://wikipedia.org if the site supports it.
This will fix a bunch of old links that are still floating around on various sites, forums, etc and keep people on https, instead of doing the https -> http -> https redirect bouncing around that can happen now.
Ah, that’s a great feature. Hope this comes to Brave soon.
I disagree. While in practice, this is often the same website, it is a different protocol and a different port. It just happens to use the same DNS address. You’re explicitly giving your browser a FQDN, and it is ignoring it and doing something else.
I hope this feature can be disabled. Google has been ignoring the W3C and has shipped proprietary, insecure features in their chromium engine for a while now, so it wouldn’t surprise me if they made it permanent 🤷
What kind of monster would deliberately serve different content for http and https versions of the same URL?
I agree. That would be absurd.
However, I don’t like not having the option of using HTTP if I want to use it. It’s okay if the webserver redirects me, but I don’t like if my browser does it when I didn’t tell it to. I might want this when doing development, port tunneling, VPN stuff, etc. In most cases, it won’t matter, but when it does, it will be a pain in the ass.
Imagine you want to test your redirect from 80 to 443 when setting up your webserver.
While I think for the normal user this enhances security by defaulting to HTTPS, however this makes no sense for a browser. This should be enforced server side, the browser is for browsing, i.e. viewing. Not controlling and competing with the server software for competency.
Chromium is really leaning into bad code practice with the disregard for “separation of concerns”.
If it’s enforced server-side, then there’s still an initial connection that is unsecured and can potentially be intercepted/modified before it gets to the redirect from 80 to 443.
I have, and for quite some time when I was trying to set Https up.
It’s really bad IMO to “decide what the user wants” even if this is both discussable and a very small step, it is a step towards that.
Pushing traffic to https isn’t the worst thing. My ask would be to have a toggle to disable due to local development or server deployments where http/port 80 is the only choice.
It does specifically say “defaulting to https:// if the site supports it”, so I think specifying http will still work if the site doesn’t actually support https.
Got a message back from Https, let’s switch!
The message:
“Internal nginx routing error.”
No testing a server side http-to-https upgrade/redirect without reconfiguring your browser. This seems like an unnecessary and bad idea.
This could be easily done better by promoting such server-side configurations as a default.
I mean, why should the browser attempt to correct inappropriately configured servers? Shouldn’t they rather be making PRs to NGINX/Apache/CAs or whatever?
Also: can’t this be exploited to spoof an unavailable HTTPS and coerce an unencrypted connection?
Not touching chromium or chrome. Sworry! Happy with Firefox.
What’s the difference between the two?
I’m not sure which thing you’re referring to.
If it’s between http and https, the s stands for secure and the connection to the server is authenticated and encrypted.
Was curious, thanks
Just to expand on that, it’s a very basic encryption, but it provides a little bit of a safe standard. When ppl talk about “encrypted communication” they usually talk about more than that. For example, apps like telegram use some more advanced encryption iirc.
The latest version of TLS (used in the latest version of HTTPS), 1.3, is very secure. Most websites these days support 1.3/128 bits, making it quite hard to crack. One major weakness of HTTPS is that, if a certificate authority is compromised, the hackers can issue certificates for ANY website, which browsers will accept as secure until the certificates are revoked/expired/CA removed from trusted list in browser. This loophole can also be exploited by nation states (forcing the CA to issue certificates).
If you are doing something really private, use something like Matrix (E2EE mode), Signal, or Telegram (E2EE DM).
TLDR: Modern HTTPS is incredibly secure, except there is a loophole that nation states and hackers can exploit if they compromise/gain control of an approved certificate authority. If you are doing something you really dont want anyone to find out (top secret files), use an encrypted service that does not rely on the TLS/SSL/HTTPS stack.
Oh, there was an effort to solve above loophole, I’m not sure if it got anywhere though.
Is there a secure option that uses all the features minus the 3rd party certificate parts?
No, they were working on a solution a while ago, where a website would list what CA it used so you couldn’t get a random CA to issue a cert, but that effort was abandoned iirc.
Their Googlebot has been pushing https hard for years. I really don’t see the point for mundane sites with no sensitive or controversial content, but there is no way to fight Google so like a good little site operator I go along if I want to be in Google search results.
