Re: HTML interpreter vs. HTML user agent

On May 26, 2009, at 12:26 PM, Larry Masinter wrote:

>> By "HTML interpreter" do you mean "HTML user agent"?
>
> The phrase "user agent"  was introduced and used in the
> literature as representing a software used by, well,
> users: humans or agents representing humans.
>
> The "User-Agent" header was introduced into the HTTP
> protocol at a time when HTTP was thought only to
> involve clients (that were user agents) and servers
> (that were not user agents.)

The reason I asked is that I was wondering if you were trying to  
identify a specific subset of HTML implementations, or just looking to  
change the terminology. Looks to me like the latter.


> "HTML interpreter" seems like a more general
> term that would include HTML user agents but also
> other kinds of agents.

The reason that the term "HTML interpreter" rubs me the wrong way is  
that "interpreter" already has a well-established meaning in the field  
of computing. Specifically, it is an implementation strategy for a  
programming language, as opposed to a "compiler". The spec doesn't  
care about the implementation strategy, and interpreter vs. compiler  
are not even categories that make sense with respect to HTML5. Having  
worked on the implementation of compilers, interpreters, and HTML  
processors, I can tell you the first two categories are not really  
relevant to the latter. So to me it sounds like it might to you if  
someone referred to an HTTP proxy as an "HTTP bridge" or "HTTP  
router". Maybe in some abstract sense it is vaguely right, but in  
reality HTTP is not a protocol to which routing or bridging in the  
canonical sense apply.

> "HTML processor" might work,
> but an "XML processor" doesn't do any of the
> semantic interpretation.

I think "HTML processor" or "HTML implementation" would be accurate  
value-neutral terms.

Regards,
Maciej

Received on Wednesday, 27 May 2009 22:03:22 UTC