Soobok Lee <[EMAIL PROTECTED]> wrote: > If a simple HTML page contains the following tag, > <a href=http://www.<ML>.com>Hello World!</a> > in which, <ML> maybe in a native legacy encoding or utf8 encoding, it > is easy to imagine that some vistors who click that link may be led to > wrong sites or nowhere.
Very easy to imagine indeed, because the HTML spec says that the href attribute must contain a URI, and the URI spec says that the host must contain only ASCII letters, digits, hyphens, and dots (or it may be a bracket-enclosed IPv6 address literal). > Should IDNA recommend all HTML authors to use such ACEed URL for > backward compatilbility and error-free fast deployment? Not necessary, since the HTML and URI specs already limit the host to ASCII letters, digits, hyphens, and dots. > Is HTTP/1.2 being planned for IDN HOST: values ? Bruce Thomson <[EMAIL PROTECTED]> replied: > Well, depending on how you want it to work, version 1.1 might be OK. > It allows %-escaped UTF-8 I believe. HTTP 1.1 says that the Host: header field must contain the <authority> part of the URI (that is, <host>[:<port>] ), and the URI spec forbids %-escapes in <authority>. > Would using ACE here be a change to the HTTP spec No. ACE host labels are honest-to-goodness valid ASCII host labels, so you can use them wherever traditional ASCII host labels are allowed. You don't need any special permission or invitation. AMC
