Btw, I agree about the session IDs in the URI (see my para above), but most of the time the request being sent isn't POST so that's not an option. If someone disables their cookies, then the application server has no choice but to attempt to propagate the session through the GET requests. (I don't think it is appropriate to make every link in a web store trigger a POST request simply to avoid putting the session ID in the URI.)

2- when people give out URLs as "visit us online at foo.com," when "foo.com" doesn't work. (This is similar to yours.)

3- Static URLs that don't end in / (that is, ones that specify a filename), but this really only bothers me if it's the main URL -see #1.

I also don't like the converse, when people refuse to use query variables, and use awkward slash-hierarchy even when it's not really part of a resource hierarchy. (Underusing it is much more widespread, because too many people fail to realize that URLs don't have to correspond directly to filesystem paths. But when they do realize this, they often misunderstand how slash-delimited hierarchy and query strings complement each other, resulting in the latter problem.)

And I'm also not fond of filename extensions. They're unnecessary, and in any good web server (Apache included) they're easy to remove, so what's the problem?

I don't like when people misuse HTTP methods. A GET request shouldn't cause a permanent change in any data or metadata about a given resource (consider bash.org's voting system for example). That's the whole point of POST, and it has practical problems also -- especially that doing things like that with GET is just asking for trouble from search spiders and pre-caching proxies. A few simple Google searches prove this: Google has voted positively on 883 bash.org quotes, negatively on 853, and has requested the deletion of 1220. And, even more inconveniently for bash.org, these are in Google's index, so they'll probably all be requested many more times. If they hadn't been lazy by making those commands be stateless GET URLs, this wouldn't be a problem.

Sometimes POST is also misused. Thankfully, this is rare, but some web designers get the idiotic idea that people should only be able to start viewing their site from the front page, and that nobody should be able to link to anything but the front page, so they program it to use navigation through forms and POST. I don't know what would drive them to commit such Web-thwarting depravities, but those designers will be lucky if I ever go back to any of their sites. This is similar to sites that use incomprehensible URLs containing session IDs, like Amazon and the Apple store, making linking hard or impossible; unfortunately, in such high-profile cases, it's not easy to avoid using such sites at one time or another.

The worst cases of this even displays two totally different sites with and without http://www... Talk about bad hosting providers!