- From: Paul Libbrecht <paul@activemath.org>
- Date: Mon, 19 Apr 2004 00:18:39 +0200
- To: Luca Padovani <lpadovan@cs.unibo.it>
- Cc: www-math@w3.org
Funnily this approach you described, which I would call "coordinate as extra attributes", is the one we use in ActiveMath. But you tackle the problem well by saying that combined markup gives a kind of self-contained-ness: coordinate-as- extra-attributes could be said to be dirty... at least it is non-standard hence, in principle, not usable by anyone else than your programs/web-pages. The self-contained-ness character is, I think, something that will not survive and I would expect applications to migrate to a model where a paste actually triggers web-requests over the web. paul On 18-Apr-04, at 08:17 Uhr, Luca Padovani wrote: > On Sun, 2004-04-18 at 00:28, Paul Libbrecht wrote: >> As I understand it, parallel markup is required if one wants to offer >> correct sub-term selection > > well, not _strictly_ required. If you generate properly grouped mathml > presentation markup it is possible that the structure of presentation > is a refinement of the structure of content. In this case, by adding > an attribute on those presentation elements that actually match a > content > subtree, you're able to discriminate between cosmetic presentation > markup and markup that "means something" without looking at the > content level. > > What mixed markup gives you is (among other things) the ability to > make the document somehow self-contained, whereas the approach that I > just described is useful if you can "trace back" the content from > presentation, perhaps looking at the value of those mentioned > attributes. Note that this is the approach used in HELM/MoWGLI.
Received on Sunday, 18 April 2004 18:19:15 UTC