4 ms·
IMHO browsers never should have gotten cookies in the first place, and they certainly shouldn't have ever allowed them to be accessed from different sites. Sure
by phkahler 11y ago
IMHO browsers never should have gotten cookies in the first place, and they certainly shouldn't have ever allowed them to be accessed from different sites. Sure, it's very convenient for a site to recognize you when you return, but there are other way to get much of the way there.
But hey, rather than scaling back HTML 5 includes a client side database! WTF?
- fixermark 11y agoOut of curiosity, what would you have recommended, had you been on the web steering committee at the time?
- phkahler 11y ago>> Out of curiosity, what would you have recommended, had you been on the web steering committee at the time? Cookies should have been at most a GUID for a single session on a single server. With a little dynamic content you can put a session GUID in every link on a site and not have anything stored locally on the users machine. Back in the day this was a more significant problem than it would be today. Note that this would also preserve the user ID in bookmarks. Also, it really begs for HTTPS. There are probably other ways to do these things, but people will either claim they are inefficient or just keep their blinders on and make excuses in favor of the status quo. Of course many of the alternatives could probably be abused as well, it's just harder.
- fixermark 11y agoThat might very well have legs. I wonder how one would support keeping a user logged in across sessions, but that might still be a solvable problem.
- rpcope1 11y agoI'm not sure why you're being downvoted, but I agree: cookies are enormous (bad) hack, and we knew they were an enormous hack when they were implemented many moons ago. I think it would be far better for HTTP to get a better auth system, and stop using cookies for auth, which would certainly mean it's far more fe. Also locking down session cookies to only get sent back to the host that generated them in the session would be excellent. It's highly unfortunate that this isn't fixed in HTTP/2, but it probably goes to show where the controlling powers interests lie.