RE: Last Call: Preferred Format for Literal IPv6 Addresses in URL's to Proposed Standard

["Larry Masinter" <[e-mail]>]

> Ah, right.  OK, then; I withdraw my first objection.  The brackets are
> still a problem, though (not one I care deeply about, but maybe someone
> who worked on 2396 will :-).

I worked on RFC 2396. The brackets [] inside URLs are not so much
a problem for the URL handling software itself (the browser "Address"
box) but rather for more specialized applications, such as software that
heuristically picks out URLs from text/plain email messages and turns
them into hyperlinks, or humans trying to read a URL in a magazine
article and type it in.

I don't have a serious problem with the URL literal syntax proposed in 

   draft-ietf-ipngwg-url-literal-02.txt

even though it isn't legal in RFC 2396. It should explicitly
'update' RFC 2396, though, rather than leave the updating implicit.

I think that the syntax in 

    draft-masinter-url-ipv6-02.txt

would cause fewer problems for URL processing software, although
it would be less problematic if the ".ipv6" pseudo-domain were
consistently implemented anywhere host names were allowed, and not
just in URLs.

The converse of "Every problem in computer science can be solved
by adding a level of indirection" is that "Every level of indirection
needs a way around." This is the case for the hostname -> address
indirection; it's very useful to have a standard form of 'hostname'
which stands for an address. But if there's consensus that it's
better to do this with "[]" brackets, so be it.

Larry
-- 
http://www.parc.xerox.com/masinter