W3C home > Mailing lists > Public > www-svg@w3.org > January 2003

Re: High-Quality Dynamic SVG Viewer

From: Tobias Reif <tobiasreif@pinkjuice.com>
Date: Mon, 20 Jan 2003 10:52:17 +0100
Message-ID: <3E2BC6D1.7070803@pinkjuice.com>
CC: www-qa@w3.org, www-svg@w3.org

Dean Jackson wrote:


> I suggest the following to resolve the issue: [...]
> - Provide explicit wording on the relationship between well-formedness
>   and streaming. At the moment, the spec says that you can only
>   render a well-formed document, which means that you can't do
>   progressive rendering (since any partially downloaded file is
>   not well-formed). The answer is to create a temporary well-formed
>   version of the partial download (close all open tags, etc) and
>   then render.


This does not sound good to me. Requiring the application to do 
guesswork will introduce lots of unforeseeable behaviour, AFAICS.
If client apps are allowed or even required to do guesswork (especially 
at such a relevant level), I might as well have a random algorithm code 
my SVGs :|

Perhaps it would make sense to not require ~"closing all open tags, 
etc", but instead specify a clear, simple, and most importantly much 
less ambiguous algorithm. Something like:

   For each subtree fragment which is completely parsed, the app can
   decide if it has enough data to start/continue rendering/animation.

Each "subtree fragment" is a well-formed XML document having one root 
element; thus, there's no guesswork required to make it well-formed.

>   IF the viewer can determine that the document
>   is truely in error (not well-formed or something else) then
>   it immediately swaps into error mode.


That sounds good.


Tobi


-- 
http://www.pinkjuice.com/
Received on Monday, 20 January 2003 04:52:50 GMT

This archive was generated by hypermail 2.3.1 : Friday, 8 March 2013 15:54:24 GMT