[rfc-i] Proposal for better lists in v3

Julian Reschke julian.reschke at gmx.de
Sun Feb 9 23:35:49 PST 2014


On 2014-02-09 23:38, Paul Hoffman wrote:
> Following up on the great thread we had recently, I have what I hope is a summary of the requirements and a simple proposal to meet them. The requirements are pretty much what Julian proposed initially, and the proposal specifics are based on Nico's outline. Comments are welcome.
>
> --Paul Hoffman
>
> Requirements:
>
> 1) List items can have more than one paragraph without triggering another list label.
>
> 2) Lists should be block elements, not inline (flow) elements. They should be usable outside <t>.
>
> 3) Give more control over the labels for "style" of "letters", "numbers", or "symbols". Make more explicit the rules for nested lists that use the defaults for a "style".
>
> 4) Make hanging lists more like other list styles.
>
> 5) Make a real "dictionary" style, with the definition starting on a new line.
>
> 6) Don't mix label style with content (the "format" attribute).
>
> 7) Create a new mechanism to cause a paragraph to be indented, without using <list style="empty">.

You chose to remove my:

5) Text content in attributes (hangText attribute) is bad.

> Proposal:
>
> <list> is changed to have each list item is represented by a <li> element. The content model for <li> is the same as that of <annotation> with the addtion of <t>, namely: text, <b>, <cref>, <eref>, <i>, <iref>, <t>, <tt>, and <xref>. Inside of an <li> element, a <t> elements causes a break but without triggering another label. The <li> element has optional attributes only where needed.

I'd rephrase that as either inline content or a limited set of 
block-level elements (just t); but either one or the other.

> <list> no longer needs to be inside of <t>, and can appear in the other contexts where a list might be expected.
>
> <list> has a new optional "anchor" attribute so that a list can be referenced from other places in the text.
>
> For the "style" attributes in <list>:
> - For "letters", "numbers", and "symbols", there is an optional "format" attribute that matches the current one for formatting the label. There will be a clear set of defaults for each of these given in the spec.
> - For "hanging", the <li> element has an optional attribute of "hangText".
> - Add a new style of "dictionary", the <li> element has an optional attribute of "definition".

I believe this makes the model worse because it adds more special cases 
(a new style-specific attribute).

> - Remove the "empty" style (to be replaced by a new <blockquote> entity).

I'd rather keep it; I've seen lists using this style. I just don't want 
to see it abused to generate things that are not lists. If everything 
that needs indentation now switches to blockquote instead of list no 
matter whether it was a quote we have made things worse

Best regards, Julian



More information about the rfc-interest mailing list