CURRENT_MEETING_REPORT_ Reported by Tom Magliery/NCSA Minutes of the MIME Content-Type for SGML Documents Working Group (MIMESGML) This group met as the SGML BOF at the 31st IETF on 6 December, and has since become the MIMESGML Working Group. Goal The primary goal of this meeting was to establish a working group to propose a standard for SGML document exchange over the Internet; i.e., to propose a MIME encapsulation for SGML (beyond ``text/sgml''). Agenda o Introduction o Statement of goals o Discussion of charter for group o Discussion of a list of issues facing the group o Creation of a calendar of milestones for the group Discussion A draft charter is being discussed by the IESG (Internet Engineering Steering Group) and may be issued the week of 12 December 94. Ed Levinson gave a brief introduction to the proposal contained in his Internet-Draft, draft-levinson-sgml-02.txt. He showed how Content-IDs are used as interchange tokens for file names in SYSTEM entities. Paul Grosso, gave a brief introduction to the parts of SGML relevant to the working group's charter. He also described the SGML Open Technical Committee Report TR 9401 which recommended the inclusion of a catalog with SGML exchange packages. A discussion ensued about providing a catalog similar to the one the SGML Open Technical Resolution 9401:1994 describes in issue B. The catalog would allow for a single place for the mapping of logical names to actual storage object locations. An SGML Open style catalog allows for the mapping of PUBLIC identifiers into their respective storage object identifiers and helps the receiving system resolve the PUBLIC identifiers. The need for the catalog was questioned and the exact MIME message format was discussed. No consensus was reached on these issues, and a suggestion was made to conduct further discussions on this topic via e-mail. Two other issues were discussed in detail, the number of parts to separate the SGML document into, as body parts in the MIME message and the MIME content-type to be used. As to the number of parts, the Internet-Draft, draft-levinson-sgml-02.txt, proposed three parts, the SGML Declaration, the prolog, and the document instance. That allowed for the possibility of SGML viewers that required those parts to be provided separately rather than as a single stream. Charles Goldfarb noted that SGML is defined as a stream and, since a default declaration exists, a single part, consisting of the prolog and instance suffices. Others pointed out that HTML, used for the World Wide Web, could likely use a similar scheme, especially upon achieving full SGML compliance. The attendees reached consensus on using two parts, not three. Tim Bernards-Lee suggested using a Multipart/Related content type instead the Multipart/SGML, as proposed in Ed Levinson' Internet-Draft. The first body part in the Multipart/Related being the beginning of the document. Discussion was deferred to allow the group to identify all the issues of concern. The issues below were identified and then categorized as being controversial, non-controversial, and not within the working group scope or charter. Ed Levinson agreed to start discussions of each controversial topic. List of Issues The following is a list of some of the issues we decided the group needs to consider. Legend: ``*'' -- Issues needing to be resolved through discussion (or strong differences of opinion exist) ``o'' -- Plain old ordinary issues (no strong differences exist) ``-'' -- Related issues that do not need to be tackled by this working group * MIME content types application/sgml versus text/sgml multipart/sgml versus multipart/mixed * Use of Content-ID Content-ID implies the receiving system may cache the body part. Should an new header be used, i.e. Content-reference? * URIs, should we used a URL style for the content-ID label (i.e., cid:...)? o Catalog Should one be explicitly included? o SDIF (SGML Document Interchange Format) Should the Proposed Standard include the section on SDIF now in the Internet-Draft? o Ordering of SGML document parts in multipart MIME parts: is it required? o Should the SGML document be transported in three parts (declaration, prolog, instance) or two, (prolog and instance combined as one). o Character sets: how to deal with documents with multiple languages and how to specify the markup character set. o Security implications Processing Instructions and NOTATION SYSTEM identifiers. - Style sheet support Only need to show how to extend work to include style sheets. - SGML fragments Sending a part of a document along with information on where in the content model parse tree it belongs. ``Swiss cheese'' or incomplete documents. Milestones/Calendar December 94: Establish an IETF working group (should happen this week). February 95 (3rd week): Internet-Draft available. March 95: SGML Open Technical Committee meetings/review of draft. April 95: Proposed Standard. May 95: Last Call on Proposed Standard. Further Information There is a mailing list appropriate for discussion of these issues. The mailing list address is sgml-internet@ebt.com. To subscribe to the list, send e-mail to majordomo@ebt.com with the body subscribe sgml-internet. There may or may not be an archive of the mailing list; Paul Grosso will be checking on this. (Even if there is not one yet, the list really has not received any traffic, so nothing has been lost.)