Minutes from the San Jose IETF meetings

Eric W. Sink (eric@spyglass.com)
Thu, 22 Dec 94 13:24:01 EST

A hearty thanks to Tom Magliery for taking the minutes!


HTML Working Group Meeting Minutes 31st IETF, San Jose, CA Thursday, 8 December 1994

The IETF HTML Working Group met at the 31st IETF in two sessions on Thursday, December 8, 1994. Tom Magliery <mag@ncsa.uiuc.edu> volunteered to serve as secretary for the meeting. Eric Sink chaired the morning session and Tim Berners-Lee the afternoon session.

MORNING SESSION ===============

Historical Information ======================

Tim Berners-Lee <timbl@w3.org> gave a brief history of the HTML specification process for the benefit of newcomers (and the interest of some old-timers).

The original specs for HTML existed only in hypertext form as part of the WWW project at CERN. Paper specifications were created from time to time but nothing was submitted to the IETF. Dan Connolly <connolly@hal.com> created an SGML declaration and DTD to describe HTML in terms of SGML. In Spring 1994, the SGML community pressed for a more rigorous specification, and Dan drew together a group to define current practice, which met in Geneva at WWW conference. At the same conference an HTML workshop defined future levels of HTML in outline. Tim encouraged the group to form an IETF WG via a BOF in Seattle. The group also met at the Chicago WWW conference. The WG charter was to define current practice (HTML 2.0) before starting work on future levels (3 and 4).

The goal for today's meeting was to wrap up the 2.0 specification, and to begin discussion of HTML 2.1 and 3.0.

HTML 2.1 is meant to be things as they are now, plus minor tweaks. (What constitutes "minor tweaks" is another matter for discussion, for example ICADD inclusions to aid Braille interpretation but which do not affect the spec otherwise are included.)

HTML 3.0 is the next generation of HTML. This includes such new features as figures and tables support. The work on HTML 3.0 is already far along thanks to the efforts of Dave Raggett.

Status of document ==================

Eric Sink <eric@spyglass.com>, one of the editors of the 2.0 document, gave a summary of the current status of the document, which was "basically done" except for minor editorial changes. The most pervasive feeling of the HTML WG was "Finish it!". In fact, the DTD portion of the document is basically set in stone, but the commentary still needs some work.

Eric Sink listed several items for discussion with regards to the 2.0 document: proposed ICADD changes to the DTD; Larry Masinter's file-upload proposal; international character set issues; what to do about end-of-lines; and "Shorten it!".

Proposed ICADD changes ----------------------

The group agreed that we should incorporate the proposed ICADD changes to the DTD, with the justification that they may not be current practice, but they are still widely agreed upon as being Good Things.

File upload proposal --------------------

Larry Masinter <masinter@parc.xerox.com> presented his file upload proposal. After some discussion of various issues, it was decided that the mere fact that there was such discussion suggested that this was a feature best left for future versions of HTML, probably 2.1.

Note: this was discussed again later in the session. See below for notes, and a URL for the proposal.

International character sets ----------------------------

More lengthy discussion resulted in the following suggested actions for HTML 2.0:

1) The baseset for HTML may be overridden by a MIME character set; 2) HTML 2.0 may be supposed to only support the ISO Latin-1 character set; 3) the charset parameter defaults to ISO Latin-1.

There are some changes needed to the text of the HTML 2.0 spec to support these ideas. Larry Masinter agreed to write a draft of these changes.

End-of-line characters ----------------------

The general problem of how and where to deal with the problems of different end-of-line characters on different platforms was discussed. Several variously feasible possibilities for dealing with the problem were mentioned: 1) change MIME; 2) define HTML as application/html; 3) [I missed this one -- mag]; 4) tag the eoln character as a special attribute of text.

A lengthy discussion ensued as to how to represent current practice in the various specs. The following list of propositions was agreed. Specs affected are in parentheses. Points marked [*] should be taken as standard behavior, and the point marked [**] should be taken as recommended behavior.

o BASESET can be overridden by the sharset attribute of text/html [*] (The definition of text/html as a MIME type) o in HTML 2.0, text/html only supports ISO Latin-1 [*] (HTML 2.0) o the MIME default text/* character set is US ASCII [*] (MIME, no change) o if you find ASCII characters > 128, then assume ISO Latin-1 [*] (HTTP, Roy Fielding to check clear in current spec) o text/* should be sent with CR/LF for eoln [*] (MIME, no change) o anyone should accept CR or LF or CR/LF for eoln [**] (HTML in general, HTTP as comment on use of MIME)

Shortening the 2.0 spec -----------------------

In the interest of reducing the overall length of the 2.0 spec, it was proposed that the DTD element references (currently section 6) be removed from the spec. It was then pointed out that some of the element references were in fact incorrect. The group agreed on the following compromise action:

1) First, check out the element references and see if they are correct; 2) if they are correct, move them to the appendix of the document; 3) if they are not correct, then get rid of them entirely.

HTML 2.0 specification wrap-up ==============================

The group decided to make the final changes proposed during this meeting, and, barring serious dissent on the HTML-WG mailing list over the next two weeks, to submit the document to the IESG for Proposed Standard status.

2.1 discussion ==============

Larry Masinter's file upload proposal was discussed again. [Secretary's note: At this point I have several notes of comments which over the course of a week, having lost their context, have become inscrutable to me. I transcribe them here and invite suggestions as to their actual meaning.]

* Allow application author control to give good hints as to what file to send. * Don't want to have default values in hidden fields, or any other fields for that matter. * (Tim B-L) Trying to do some independent things together, 1) Allowing people to send attachments as opposed to inline field data; 2) Specifying Accept-able types fopr forms 3) Sepcifying a new form-data type. The principle aim of the proposal is (1). (2) is also useful but should be orthogonal. (3) may be inferior to using HTML for the return data from the form.

The URL for the most recent (as of this writing) version of the proposal is:


AFTERNOON SESSION =================

HTML 3.0 ========

Dave Raggett <dsr@hplb.hpl.hp.com> gave a summary of his work on the initial draft of HTML 3.0. Some of the new features included in this work are resizable tables, figures with local event processing, mathematical formulas, document specific toolbars and client side form scripts. Dave demonstrated many features as implemented in his "Arena" browser.

At present there is no specification document, but the draft DTD, which is quite heavily annotated, is available at the following URL:


One important point is that HTML 3.0 should be backwards-compatible with the 2.0 specification. So far, it is.

Another hot issue is presentation control; style sheets are one solution being discussed (URL later). Another approach is to embed some form of presentation control into HTML. There are several ways to do this: 1) add new elements (considered Very Bad); 2) use SGML processing instructions; 3) add attributes to existing elements. An initial attempt at this last approach has been done with HTML 3.0. See the list elements in the DTD for examples.

Further discussion took place, mostly concerning features of table and figures in the 3.0 draft DTD.

Netscape things not already in 3.0 ==================================

Tim Berners-Lee made a quick pass through Netscape Communications' documentation of their extensions to HTML, in the absence of a representative from Netscape, to see what additions they have that are not already in the 3.0 draft DTD. We came up with the following list of tags/attributes:

o <CENTER> o <ISINDEX> -- PROMPT attribute o <HR> -- lots of new attributes o <UL> -- TYPE attribute o <IMG> -- major changes o <BR>, <NOBR>, <WBR> o <FONT> -- all kinds of stuff o <BLINK> (not actually documented in their spec?)

Due to limited time, discussion of these elements was minimal. The Netscape HTML extensions may be found at the following URL:


Another <FONT> proposal =======================

Alex Hopmann <> presented a proposal for a <FONT> tag with capabilities similar to that of Netscape's. A short religious war ensued, without any real resolution. The needs were voiced for

1) Inline presentation markup; 2) Powerful style sheets 3) Consistency between 1 and 2.

DSSSL Lite information ======================

In the last couple of minutes of the meeting someone expressed interest in the location of information about the work that is being done on "DSSSL Lite", a style sheets proposal for HTML. This and other style sheet work, collected by Hakon Lie <howcome@dxcern.cern.ch>, may be found at the following URL:


Here endeth the minutes =======================

Respectfully submitted, Tom Magliery <mag@ncsa.uiuc.edu>

Eric W. Sink, Senior Software Engineer --  eric@spyglass.com
                                           I don't speak for Spyglass.
"Can I get a direct flight back to reality, or do I have to change planes
in Denver?" - The Santa Clause