[rfc-i] Normalized HTML
masinter at adobe.com
Wed Mar 28 01:16:04 PDT 2012
After listening to the BOF discussions and reading the email, I've changed my thinking.
I'm leaning toward a marked up profile of HTML where the authoring format and distribution format is a canonicalization of the authoring format. That is, I think we could move away from XML/XML2RFC and instead have a "cleaned profile" HTML with a tidy-step that does the work that xml2rfc does but which is idempotent (i.e., the output is suitable for input and generates the same HTML.)
Having the output format suitable for input and production of other formats trivializes the re-use issues.
"View Source" is powerful.
ID-editor: edit in simplified profile of HTML, or else edit in some tool and generate ("clean") HTML.
Preprocessor run through a 'tidy' which does things like "make head/title match h1 header", fixing up metadata, annotating or styling references, fix cross-references.
Postprocessor: generate ASCII-only view of HTML by substituting longdesc for diagrams, changing cross-references, etc.
Reviewers can comment on HTML or derived forms (ASCII-text view, PDF, epub, etc.)
Validate ID-submitted form, edit in same way that ID-editor does, if necessary.
Convert RFCs also into .txt, .pdf, epub.
If they appear, diagrams MUST be in SVG. Equations MUST be in MathML. No images allowed(!) (Text should be text so it is searchable). Probably need some "clean" styling requirements for diagrams.
Equations and diagrams must have ASCII-only alt descriptions for accessibility, meet accessibility guidelines.
Probably something similar to W3C pub-rules and using tools similar to W3C pub tools.
More information about the rfc-interest