[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: I-D ACTION:draft-fenner-literal-zone-00.txt



How interesting! www.%33com.com also fails, so it isn't even
being kicked into numeric addresses that does it - the parsers
simply don't interpret the escape at that point.

So they're all broken?

    Brian

Juergen Schoenwaelder wrote:
On Sun, Dec 05, 2004 at 11:54:07AM -0800, Bill Fenner wrote:

I've added the following pieces to the draft to try to capture the
conversation so far.

[at the end of section 2.1:]
  o  Use '%' in the URI
     Pro:
        +  "%" is the same character.
        +  Can copy and paste between forms.
     Con:
        +  '%' is fundamentally special in URIs; parsers can be
           expected to be hard-wired to know that they start a
           percent-encoded octet.
        +  IPvFuture ABNF doesn't permit bare percent.
     Issues:
        +  Impossible to ensure that this exception to a fundamental
           URI rule would be handled properly by parsers.


I think it is useful to explore how generic these real-world URI
parsers really are. I just tried to use

	http://%3134.169.34.18/	 and  http://134.169.34.%31%38/

to access a web page, expecting that parsers would turn that into

	http://134.169.34.18/

but I got lots of "host not found" messages. So it indeed looks like
the real-world parsers I have access to pass hostnames/IP addresses
to name lookup functions without preparsing them. I also tried things
such as <http://%77%77%77.ietf.org/> without much luck. Please make
your own tests.

The conclusion from my little series of tests is clear. The
implementations I have tried do not interpret % escape sequences
in the host portion of URIs and so the direction should be to
simply allow % for IPv6 addresses with a zone identifier and to
tell URI folks that they better do not mandate that % escape
encoding is applied to the host portion of URIs since this is
a) not really useful and b) not widely implemented. Of course,
if you find implementations that do interpret % escape sequences
in the host portion of URIs, your conclusions might be different.

/js


-------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@xxxxxxxx Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------