- From: Chris Lilley <chris@w3.org>
- Date: Tue, 10 Jan 2006 19:59:28 +0100
- To: www-svg@w3.org
- Cc: T Rowley <tor@cs.brown.edu>
Hello www-svg, Tor writes > The allowed attributes for <prefetch> allow a very detailed description > of the prefetch. I have very little confidence that authors or content > generation tools will be able to usefully set these, as generally they > won't be aware of the memory, network, and other conditions that will be > encountered when the content in being loaded. This will likely wind up > in implementations ignoring the detail and using the prefetch element as > a boolean hint. Why not just simplify the element to that level now? There is no dependency on memory in setting these, so content creators do not need to be aware of that. Instead, you typically just say how big the thing is, which can be determined (by the authoring tool, or as a post-processing step) by doing a HEAD request on it. In most cases, content is served in a single resource variant, but we need to allow for the case where it isn't. Thus, its a hint with attributes to detect staleness (ie a different variant is being served, or more likely the resource has changed). In most cases the content is binary, but for the cases where it is textual we do allow for the encoding (eg UTF-8) that was used to measure it to be specified. Again, most content is served using only a single encoding; the attribute can be safely omitted unless content happens to be served at the same URI in multiple encodings, which does happen. It should be clear which attributes are optional and what the defaults are if they are omitted and what cases need them - is it? So a typical use - to get some video, for example, is <prefetch mediaSize="1234567"/> That is not particularly complicated. We do allow for setting network effects, too (as SMIL does). Otherwise, people worry that prefetch will use up all the bandwidth and impair interactive performance. I agree that finding a suitable value is not easy here, but that is an attribute we inherited from SMIL. -- Chris Lilley mailto:chris@w3.org Chair, W3C SVG Working Group W3C Graphics Activity Lead Co-Chair, W3C Hypertext CG
Received on Tuesday, 10 January 2006 18:59:32 UTC