It’s a nerd story about short timeouts. Effectively a what is the speed of light or electricity in copper and over infrastructure. It’s a joke that doesn’t make any sense because 3ms was clearly bullshit devised for example. Don’t think about it too hard, it doesn’t suddenly snap into anything meaningful.
connect() will take time. Either you then fail on reiceiving EINPROGRESS, or you attempt a select() with 0 for the timeout, which will also take time. That that time could add up to 3ms on a mid-90's system also used for other things seems entirely plausible to me.
Data can only go about 500 miles in 3ms, and in the original story, that's how long the system took to time out, and would fail to send the email
It’s a test of an old probably apocryphal story about a university that couldn’t send an email more than 500 miles: https://web.mit.edu/jemorris/humor/500-miles
Both the person who supposedly configured Sendmail, and the person who wrote the story, have defended the truthfulness of it on HN in the past.
Good to know! I do love the story.
Commenting late on this but I seriously didn't know about the story beforehand.
It’s a nerd story about short timeouts. Effectively a what is the speed of light or electricity in copper and over infrastructure. It’s a joke that doesn’t make any sense because 3ms was clearly bullshit devised for example. Don’t think about it too hard, it doesn’t suddenly snap into anything meaningful.
Why do you think it's "clearly bullshit"?
connect() will take time. Either you then fail on reiceiving EINPROGRESS, or you attempt a select() with 0 for the timeout, which will also take time. That that time could add up to 3ms on a mid-90's system also used for other things seems entirely plausible to me.
[dead]