Hypertext Markup Language BOF (HTML)

Reported by Tim Berners-Lee/CERN


Introduction

A BOF was held to determine if a HyperText Markup Language Working Group
should be formed.  There were 59 attendees.  Dan Connolly and David
Ragget had expressed regret at not being able to attend.

Paper documents distributed at the meeting were the draft charter of the
proposed working group and the current (of July 21, dated July 31!)
version of the HTML specification.


Presentation

The chairman stated that the purpose of the meeting was to consider the
creation of a new working group to work on the HyperText MarkUp Language
(HTML) specification.  He began the meeting with the history of HTML:


   o The original HTML specification was kept on the web only, and
     edited by the author in response to discussion on the net.

   o Much work was later put in by Dan Connolly.  He cleared up its
     relationship to SGML and tested it.

   o An Internet-Draft was produced.

   o The draft was discussed in the IIIR Working Group.  There was much
     deliberation on the fact that it was an existing practice not
     designed within the IETF. The alternatives of Informational status
     or Proposed Standards status were discussed without resolution.
     The matter was put to the area directors who proposed it be put to
     the IESG. They did not come up with an answer.

   o The Internet-Draft expired.

   o David Ragget put in a lot of editorial work on the document and
     included a lot of the new features which were in common discussion
     on the net.  It was not clear whether the result of this should in
     fact be a rigid superset of HTML so, in the meantime, it was known
     as HTMLPlus.

   o A group of vendors wishing for WorldWide Web compatible tools
     demanded a quotable specification for HTML. An informal HTML
     Implementors Group (HTML-IG) was formed.

   o HTML-IG met during the WWW94 conference in Geneva, which also
     contained an HTML workshop.  This produced a useful understanding
     of the feature sets into which new developments could be grouped.

   o Paul Mockapetris, IETF Chair, encouraged the pursuance of WWW
     standards in the IETF context and, at the last moment, the present
     BOF was organized.

   o Karen Mulrow, colleague of Dan Connolly, edited the HTML
     specification of existing practice to increase its legibility and
     produced the draft which was distributed at the meeting.

   o An informal meeting of HTML-IG was held on 25 July at the offices
     of SCO, Toronto.  A few technical but mostly editorial changes were
     proposed to the specification of existing practice.  The essential
     points of a draft working group charter were drawn up.


Proposed Charter

The focus will be to:


   o Describe new features before developing them
   o Base the specification on existing practice
   o Express the relationship of HTML to URI, MIME, SGML and HTTP
   o Define conformance levels
   o Define transition possibilities and compatibilities between
     versions and levels


Levels of HTML

The conclusion of the WWW94 workshop was that the features could be
grouped into levels such that software capable of implementing level N
may reasonably be required to implement levels k < N.  Therefore, a
level number rather than a bit mask used to represent application
capabilities.  The levels proposed were as follows:


   o Level 0 - Mandatory.  Headings, lists, anchors, etc.  (Provides the
     least differences in presentation between platforms.)

   o Level 1 - Images and emphasis.  (Can be confusing if used for
     semantics not evident also from text.)

   o Level 2 - Forms.  (Requires greater implementation effort.)

   o Level 3 - Not an existing practice.  Tables, figures and other
     effects breaking normal text flow.

   o Level 4 - Mathematical formulae.


Levels 0, 1 and 2 have multiple implementations.  Levels 3 and 4 are
experimental as yet.


Discussion

The IESG now has a policy for incorporating existing practice on the
IETF standards track.  It will see whether the issues addressed had
wider implications than the existing applications, and if so, would
broaden the discussion.

The specification should describe WWW's use of relative URIs, as they
were not defined in the URL specification (which was in a final stage of
discussion in the morning).

A new mailing list has just been created which has around 300 members
after three days.  (Subscription requests should be sent to
www-security-request@ns1.rutgers.edu.)  If the kernel of interested
people who are willing to work together can be found, then a working
group on the issues would be useful.  There are two independent
implementations of secure additions to HTTP.

The working group charter should state that the specification to be
produced should mention the relationship (for example the mapping
between equivalent concepts) between HTML and HyTime.

The charter should be as clear as the transparency about the guidelines
for the working group.


Conclusion

The BOF unanimously proposed that an IETF working group be created under
the charter as amended to reflect the comments made where appropriate.

Ed Levinson mentioned an effort underway to investigate the conventions
for distributing SGML as a MIME type.  It was pointed out that this
involves resolution of identifiers used in references, and therefore has
an impact on URIs.  A mailing list exists for the discussion, and a BOF
is proposed for the next IETF. To subscribe to the list, send mail to
mime-sgml-request@infoods.unu.edu.