CURRENT_MEETING_REPORT_ Reported by Tim Berners-Lee/MIT - W3C Minutes of the HyperText Markup Language Working Group (HTML) Agenda o Charter o Tables o META element o Superscript and subscripts (SUP-SUB) o Internationalization (I18n) Charter The charter is very out of date. The past milestones have, where relevant, passed. New milestones are needed, such as in May the group decided to tackle remaining HTML Level 3 and 4 extensions as separate documents. Milestones will be set as extensions are taken on. The issue of MIME-type registration cannot be settled until the BOF at 19:30 about MIME-type registration in general. Tables A draft was released just before the deadline for this meeting. Few had read it. None of those who had read it had problems with it. Two people felt that they might have comments if they were to read it. The draft will be left for review on the net for 30 days and then be sent to the IESG for Last Call as a Proposed Standard. META Tag, Etc. Stu Weibel/OCLC presented the current problem. META elements had been proposed within the HEAD element, but many current browsers do not ignore contents of unknown elements within HEAD. Stu presented requirements that old browsers should ignore metadata, and that new browsers should be able to parse its complex nested structure. He proposed three alternatives: 4.1 Fix HEAD so that browsers must ignore content of unknown elements within it. 4.2 Leave HEAD ``broken'' and introduce a new CITATION tag with the property that unknown tags within CITATION would be ignored. 4.3 Fix ``META'' so that unknown META content is ignored. Another proposal was to move the metadata out of the HTML altogether into another part of a multipart (LM) or into another linked document (TBL). The problem of legacy browers putting metadata information on the screen does not do away with 4.1 to 4.3. One could use SGML attributes but the problem then is that SGML restricts the value of the attribute to limit for example nested elements within it, unless one used a non-SGML syntax. An extendible tagset was required for metadata, with a DTD for each, but this would require a browser which did not understand the DTD in question to use some heuristic above SGML to know how to skip the metadata. The metadata fields have been in discussion and flux for so long it would seem sensible to isolate their specification from the HTML specification. Using a multipart part or a linked document would allow the metadata language to vary independently of the data language. Another suggestion was to retrieve metadata with a META method in HTTP. Would this be too complicated? Without some concrete proposals, it is difficult to discuss this in detail. There is a real proposal for metadata syntax in the URC group. There is a solution to resolve the unknown HEAD element but it involves look-ahead. Stu Weibel will write up a draft by 31 August. SUP-SUB The current consensus allows no nested SUP and SUB but does allow anchor and highlighting within a SUP or SUB. Stu Weibel will put this consensus into a very short draft by 31 August. New Agenda Item: Process, Document Version Naming The consensus was that changes to the specification should be small standards and that ``glue'' documents defining new standard levels in terms of those RFCs would be released from time to time. Internationalization Currently Netscape and Spyglass' Enhanced Mosaic will accept different character encodings. They use an user-overridable heuristic to guess the character set, but will work properly if they see a charset= parameter on the MIME content-type. No one in the room was doing anything to further any specification writing here, and previous volunteers had gone quiet. Dan Connolly volunteered to see to the writing of a specification for the charset parameter, by delegation or action as need be, by December. Language tags were a different issue. They should not be lumped together with ``charset'' or SUP/SUB as one would hold the other back. Some work on language tags has been done and should be disentered. There is a problem with Unicode that it uses the same code to represent different glyphs as a function of the language being Chinese or Korean. EBT solve this problem by regarding it as a font choice issue in a style sheet. No one else had a solution. Dirk-Willem van Gulik/Centre for Earth Observation of the EC offers a database of multinational documents as test data to developers in the group. Any Other Business There is a draft (*eastlake*) on putting information about payment into HTML, etc.