If you read his page, you'll see that he's a scrambler ("peak bagger"), not a technical climber.
It's truly surprising how many programmers and engineers are into climbing of some sort (given the proper geographical locale). Rock climbing and writing code are practically all I do...
Here's why: they break TCP.
IE sends out a RST instead of an ACK|FIN to close a connection. No big deal, but it's against the spirit of the spec, which states RSTs are for emergency-type conditions.
More seriously, IE and IIS manipulate sequence numbers when they talk to each other, in order to hold open connections and not create new ones. Basically one side sends over a packet (often just an ACK) with a sequence number greater than the other side expected. This is the signal that a new "connection" has begun; I guess this would be good for pipelined requests and so on. But it breaks the spec. I'm not sure where their support for this stuff is; I would not be surprised if it was at the kernel level, thus lending credence to their claims of integrating IE and the operating system. But it hardly makes good evidence in court.
Slashdot Mirror
If you read his page, you'll see that he's a scrambler ("peak bagger"), not a technical climber. It's truly surprising how many programmers and engineers are into climbing of some sort (given the proper geographical locale). Rock climbing and writing code are practically all I do...
Here's why: they break TCP. IE sends out a RST instead of an ACK|FIN to close a connection. No big deal, but it's against the spirit of the spec, which states RSTs are for emergency-type conditions. More seriously, IE and IIS manipulate sequence numbers when they talk to each other, in order to hold open connections and not create new ones. Basically one side sends over a packet (often just an ACK) with a sequence number greater than the other side expected. This is the signal that a new "connection" has begun; I guess this would be good for pipelined requests and so on. But it breaks the spec. I'm not sure where their support for this stuff is; I would not be surprised if it was at the kernel level, thus lending credence to their claims of integrating IE and the operating system. But it hardly makes good evidence in court.