On Thu, Apr 29, 2010 at 2:47 AM, Adrian Sutton <[email protected]> wrote: > This is certainly true, but it lends support for both the old and the new > models working in tandem. Essentially there are two key use-cases for > content management systems: > > 1. Authors being able to easily create content items. As you note above, > h1-h6 makes far more sense to authors than invisible section elements and > are far easier to apply correctly. Essentially content authors expect things > to work roughly like Microsoft Word which the old model of headings does and > so they generally get it right (especially if you disable the ability to add > inline font size styles so they can't build a heading look-a-like manually). > > 2. The CMS needs to be able to assemble various content items in fairly > arbitrary ways but still get a sensible heading structure in the result. > The section element and the new heading model is an ideal solution for this > case. > > A planet aggregator is a good example of this in action - each blog post is > an entity to itself so would use headings starting at H1, but when > aggregated the headings should start at H2 (or 3 or 4 depending on what else > is on the aggregated page). So the blog author writes the post using H1-H6 > because it's easy (or section if they prefer) and the aggregator simply > wraps each blog post in a <section> so the resulting heading structure still > works.
Correct, the aggregation use-case was one of the big reasons for the heading-level scoping. > As I understand it, while the spec recommends using only H1 if you're using > sections, it's still valid and well defined to use h1-h6 along with section. Exactly. I certainly use both - sections starting at <h1> within the page template, and <h1>-<h6> in page content, because the former is nesting-agnostic, but the latter is easy to generate with Markdown. ~TJ
