- From: Peter Sorotokin <psorotok@adobe.com>
- Date: Wed, 03 Nov 2004 13:41:43 -0800
- To: Ian Hickson <ian@hixie.ch>, Robin Berjon <robin.berjon@expway.fr>
- Cc: Håkon Wium Lie <howcome@opera.com>, www-svg@w3.org
At 04:22 PM 11/3/2004 +0000, Ian Hickson wrote: >On Tue, 2 Nov 2004, Robin Berjon wrote: > > > > Ever seen poetry laid out inside a shape? Ever seen ad text following > > the shiny curves of the latest spacecraft? Ever seen some sombre lament > > about the passing of time animated as it falls through an hourglass? > > *That* is what it's for. It's for text when used as graphics. > >All three of those examples are great examplies of documents that need >semantic markup. Might - or might not. What is your point: that the feature can be abused or that the feature _never_ does the right thing? I gave (earlier in this thread), these three examples: maps, schematics and diagrams. Just look at basically any printed map and you'll see a lot of line breaks (definitely *every* one of my AAA maps uses multi-line labels a lot). Complex diagrams and schematics also use them quite a bit. What is even more important in all of the above cases is that *text is in the shape*. It might not even have word wrapping there, it might be, in fact, a single word, but it is extremely important that it is simply *placed* in the shape. Yet another example is almost anything produced in modern illustration authoring tools, such as Illustrator or CorellDraw. Text in these drawings cannot be accurately captured in SVG 1.1. > Sure, they are presented with lovely shapes. But at the >heart of the issue, they are still text, and it would make just as much >sense for them to be rendered aurally using a speech CSS stylesheet, or to >a TTY using a UA's built-in styling rules, or to have them indexed using >Semantic Web inference rules. > >If those three examples are examples of when multiline text is to be used >in SVG, then multiline text in SVG should be done by applying SVG to >documents in other markup languages, not by adding more text markup to >SVG, in clear violation of both AWWW and WCAG. Again, according to this (quite radical I should say) argument, SVG must not have any text whatsoever. Text is still text, even if it is only a single word. It does not matter for anything above that text is multiline. Along the same lines as above, one could even easily argue that SVG itself should have happened, as fundamentally, anything that SVG has lacks semantics: there should not be an svg:circle element, but there should be, say, cosmos:sun and cosmos:earth. If you disagree with a need for purely presentational languages in principle (and SVG is presentational language, XSL:FO is another and SMIL to some extent too), than there is little point in arguing about particular features of such languages. Right now people position diagram/map/schematics/illustration labels absolutely. You are saying that if we give them ability to do exactly the same thing, but express a connection between the label and the shape, so that tools and UAs can actually extract that connection, that will somehow worsen accessibility? I think it will improve it. Again, I am all for the integration with CSS flows in the future, but it should not prevent SVG 1.2 from adding purely presentational, graphics layout features to the language, even if it has to do with text. Peter >-- >Ian Hickson U+1047E )\._.,--....,'``. fL >http://ln.hixie.ch/ U+263A /, _.. \ _\ ;`._ ,. >Things that are impossible just take longer. `._.-(,_..'--(,_..'`-.;.'
Received on Wednesday, 3 November 2004 21:42:58 UTC