RE: HTML interpreter vs. HTML user agent

On Tue, 26 May 2009 21:26:14 +0200, Larry Masinter <masinter@adobe.com> wrote:
> When spiders, search engines, proxies, and other
> web intermediaries got added to the web architecture,
> the header "User-Agent" remained, even when some of
> those agents are not operating directly in service
> of a human end user.

And Anne replied:

> I have the feeling the term "user agent" remained as well.
> At least within the W3C. So I'm not sure this distinction
> makes sense here.

If the only kind of HTML interpreter that matters here
is a HTML User Agent, then the distinction would make
sense here.

However, there has been strong push back that the
specification being reviewed should be applicable
to all kinds of HTML interpreters, and not just
HTML User Agents.

There are likely important distinctions between the 
operational requirements of HTML interpreting agents
being run by a user in order to access the information
in the HTML page (and thus a HTML User Agent) vs.
those being run as a service or for some other
purpose (e.g., search engine spiders, for example).

The operational requirements would include the
requirements for security or privacy: it might
be reasonable to ask that a user agent present
the user with a security dialog asking the user's
intent or warning the user, but unreasonable to
ask that someone (a person) be presented with 
that dialog if the HTML interpeting agent is
a service scanning the Internet or even a
particular site.

The operational requirements of different kinds
of HTML interpreting agents are sufficiently
different that it is important to distinguish
between them when reading and reviewing the text,
in order to validate that the specification
in question simultaneously meets the operational
requirements.

If W3C issues a technical specification of the HTML
language, it would be expected that the language
definition would be useful by a wide category of
HTML creation and interpreting agents, not just a
specification of HTML User Agents.  The document
purports to address a wide variety of HTML interpreting
contexts, and so I think it is useful to keep in mind,
and distinguish between, the categories of interpreting
agents while reviewing the specification.

Using language -- different terms -- in mind while
doing that review -- is one way of making the
distinction and keeping it in mind.

It does seem that the largest bulk of review is being
done by implementors of HTML User Agents. But being
clear that "HTML interpreter" is a broader category 
than "HTML User Agent" makes sense.

Sincerely,

Larry
--
http://larry.masinter.net

Received on Wednesday, 27 May 2009 15:50:04 UTC