News for hackers. Stuff that matters.
Re:HTTP/1.1 Must Die by AndyKelley on Friday July 25, 2025 @02:51PM (#44686821)
HTTP/1.1 is nice and simple compared to HTTP/2, which I think is severely overengineered.

This "desync" attack seems completely pointless, it's not going to bypass TLS, and if you're not using an encrypted transport layer, HTTP is the least of your concerns...

Re:HTTP/1.1 Must Die by gruntz on Friday July 25, 2025 @03:08PM (#44687063)
Actually TLS does not protect you here. Problem is that reverse proxies do reuse backend connections and single backend connection may deliver requests from different users.
Re:HTTP/1.1 Must Die by AndyKelley on Friday July 25, 2025 @11:14PM (#44691020)
Sounds like not HTTP 1.1's problem.
Re:HTTP/1.1 Must Die by jsnell on Saturday July 26, 2025 @11:13AM (#44694713)
If a protocol is getting implemented wrong repeatedly and independently, what is the value in it being nominally "nice and simple"?

You might not be understanding just how complex the H1.1 feature set really is, and how how much fragility comes from that complexity needing to be retrofit to a too-simple design later.