Re: Advice on making IRI document suitable for reference by HTML (and other specs)

> I told you before, the solution to this problem is to not use
> the same term for both the input and output of the algorithm.

I agree that the input and output are important. For better
interoperability, two implementations, when given the same input,
should produce the same output(s). When the input is an HTML <a
href="http://...">, there are several possible outputs, depending on
further inputs such as mouse clicks and JavaScript:

DNS NAME
TCP Destination Port
HTTP Request-URI, Host
DOM HTMLAnchorElement href, protocol, host, hostname, port, pathname,
search, hash
Browser URL field, status bar
Search Engine href, display URL

Since the IRIbis work is likely to take a long time, my recommendation
would be for the HTML5 drafts and "final" spec to always have a
complete set of rules and recommendations for URL processing, with
references to other drafts/specs such as URI, IRI, IRIbis, HTTP,
HTTPbis, and DOM, as appropriate. It may be that some HTML
implementation behavior details can eventually be pushed down to
IRIbis or even URIbis, but there is no need to wait for those working
groups to reach rough consensus. (Of course, the HTML WG might take
long to reach consensus too. :-)

Now the terminology. "Web reference" doesn't quite seem right since
FTP and SMTP predate the Web. I think terms like URL, absolute URL and
relative URL are fine. We just need to remember (and document) that
the rules and recommendations are different, depending on the context
(e.g. HTML href vs rel, producer vs consumer).

Erik

Received on Thursday, 31 December 2009 16:11:34 UTC