- From: Al Gilman <asgilman@access.digex.net>
- Date: Thu, 23 Apr 1998 12:39:53 -0400 (EDT)
- To: phdavidl@usc.es (David Suarez De Lis)
- Cc: w3c-wai-ig@w3.org
to follow up on what David Suarez De Lis said: > > > > I am undergoing some experiments on tipology of links on HTML. The work is > > still on its undies and I am just experiementing with some solutions taken > > from the Web and/or Hypertext authors/sites. > (I have yet to really learn all the stuff regarding HTML 4.0 and CSS for > multiple language versions... :) I think that you will find working with HTML4 and CSS2 allows a more gracious implementation of your ideas, although some of what you want to do may require DSSSL and not CSS2. Examples: [David] X-URL: http://web.usc.es/~phdavidl/pruebas/hypertextWAI-tidy-en.html There are basicly only few different kind of links (references or relations) which are of concern for the average Web-naut, which can be classified among other criteries: 1. Semantically: 1. links to bibliographic references (or similar) 2. links to references to the same text 3. links explaining something ("footnotes", &c...) 4. links to indices 2. Location: 1. Current: same document 2. Local: same web site 3. Remote: other sites 3. Functionality: 1. file retrieving (ftp, gopher, http to a non legible document...) 2. remote conection (telnet, news, http, &c...) [secure or not] 3. comunication(mailto:, forms, &c...) 4. "Next" 5. "Previous" 6. Go to index, table of contents, &c... 7. Go to help 8. &c... [Al] This information is of marginal interest, not central. One would wish to convey it unobtrusively. Styles in a rich graphical environment are a good way to handle this sort of subtle modulation. Unfortunately, flat ASCII text doesn't offer as many subtle gradations by which to denote marginal class-shifts in text or links. The "orthodox" approach based on the model WAI-PF has been using in reviewing HTML4 and CSS2 is that you would spell out any classification not obvious from the required markup in REL= and CLASS= values (which should be mnemonic, not single letters or opaque codes), and key styles to those so that color, font, font effects such as italics, and text effects such as bracket choice (among [, {, <, etc.) can be used to convey these distinctions in different contexts. The fact is that the post-fix (a) style of representation is a bit of a pain in speech, and a prefix classifier that is a whole word (analogous to what pwWebSpeak does in announcing links) is more desirable. So adaptive styling of this connotation is desirable, with a concept name used in CLASS markup in the HTML as the cross-medium method of capturing the link class information. There is one gap in capability between what you want to do and CSS2, let me warn you. Your "functionality" distinctions can be inferred from the URL scheme in the attribute values, and perhaps a file suffix in the URL as well. Style selectors in CSS2 don't get into regexp matching on attribute values, so you may wish to play with DSSSL to experiment with your presentation concepts with a large corpus of web content. [David] In order for a HTML link to present information regarding itself (ie. the metainfornation associated to that message, which reports on the type of relation between the current document and the target document or node) we will have to add it by hand in such a way that it is the cleanest way, allowing thus shorter (but meaningful) link descriptions, but at the same time in a way that is not very intrusive (or there would not be any gain in adding information at the cost of losing legibility) [Al] You may find that, if the author has to do it by hand, they will sooner come to an understanding of how to tell people in natural language in the surrounding text than they will embrace you class codes and use them right. I don't know which will be easier to get people to do. The general topic is very interesting. You are bumping up against two tradeoffs: - does the reader want to be told about nice distinctions in semantic shading, if there is no gradual way to denote it? [varies with display medium] - the "function" distinctions relate both to what is in the markup but also what is in the verb repertory of the client software. Whether an href is a navigation or download option depends on the client system configuration. The author doesn't in general know. So you will get into contention over definition of the "verb classification" between the browser and the author. [varies with browser design and plug-in population]. I think you will find that HTML4 and CSS2 give you capabilities that you will want to align your recommendations with before you are done, but that if you experiment with these ideas you may well learn things about the effectiveness of this or that style that we don't know. I would suggest that you look at tools such as Jade that will let you write HTML4 and stylesheets and automatically down-convert this into hard-coded alternative pages in HTML3.2. This will give you a means of evaluating styling strategies in advance of the availability of the necessary HTML and CSS features in browsers. Al Gilman
Received on Thursday, 23 April 1998 12:46:50 UTC