- From: Shane McCarron <shane@aptest.com>
- Date: Fri, 23 Jan 2009 08:42:33 -0600
- To: XHTML WG <public-xhtml2@w3.org>
In speaking with Mark yesterday, I realized that the strategy we have of bundling our schema implementation(s) with our specifications might be making it harder for our readers. The specifications are getting so large that they are becoming unapproachable (the M12N 2 spec is 300 printed pages, but that includes our issues and schema). I wonder if we should consider splitting these specifications? Part 1 could contain the prose and abstract module definitions, and Part 2 could contain the implementations? Would that make the specs less intimidating? Would it make it easier for us to update the implementations as we find errors in the future? On a related note, we will soon need to decide on what is in the XHTML 2 spec vs. what is in the XHTML M12N 2 spec. Right now I am just duplicating all the content they have in common, with the M12N version saying that the XHTML 2 version is definitive. That's just to avoid confusion in the short term. In the long term, I believe we should adopt the following strategy: 1. Leave the independent modules we have defined free-standing if they can be (XHTML Access, Role, RDFa). If they cannot, either update the independent version or migrate them into M12N 2.0. 2. Have M12N 2.0 be the definitive collection of core modules and their semantics. Include in this document not just the modules used in XHTML 2.0, but also any legacy modules that we want to bring forward (legacy forms, legacy presentation). 3. Have XHTML 2.0 contain the complete (not minimal) content model definition for each of the modules it uses, but contain no semantics - instead deferring to M12N or the free-standing specifications for those definitions. I think that this approach is most consistent with our charter and our goals. Moreover, it gives our constituents the greatest flexibility in constructing new markup languages using XHTML Modules and modules from other sources (XML Events, XForms, SVG, MathML, ARIA). On the downside, it means that the XHTML 2.0 specification is not internally complete - it will not contain all of the information a document author needs to understand what each element / attribute does. However, I think we have already crossed this bridge, since many of the modules on which we depend come from other sources and we are not duplicating their content (for obvious reasons). The other disadvantage is that we are again producing a specification (M12N 2.0) that is a toolkit, not a markup language. This seems to confuse some people. However, I think we have a long track record of producing such tools and we should continue to give our constituents the tools they need to do their work (e.g., Jabber, DAISY). Opinions? -- Shane P. McCarron Phone: +1 763 786-8160 x120 Managing Director Fax: +1 763 786-8180 ApTest Minnesota Inet: shane@aptest.com
Received on Friday, 23 January 2009 14:43:13 UTC