Re: Missing Resources Indicator

Alexander Adam:
> > Shouldn't it be possible to declare a standard "icon" in the svg
> > spec to be used for missing references? As of now, Renesis v3 will
> > show an explicit marker (defined as SVG;) when something is missing
> > whereas other implementations (most of them) do actually behave as
> > if there've been no fill / stroke for example, not sure what they
> > do for images (Erik?). This might lead the author to be pretty much
> > confused at the very end about what's going though as I am not sure
> > whether he might really be able to automatically know that a no-fill
> > / no-stroke might mean an invalid-reference thingy? Just my $0.02

Doug Schepers:
> Actually, we went pretty far down the line to defining this in SVGT1.2  
> (I wrote up quite detailed spec prose for it, including allowance for  
> fallback behavior), but we decided to back out... I don't recall why  
> offhand.  I'd like to specify this in SVG2, if there aren't any  
> blockers.  It would certainly clarify what's going wrong, for users.

How about something like the following (which I think is what Alex might
have in mind):

  <style>
    image {
      image-fallback: url(#blah);
    }
  </style>
  <defs>
    <g id='blah'>
      <rect width='100%' height='100%' fill='red'/>
      <text x='50%' y='50%' text-anchor='middle' font-size='12px'
            fill='white'>broken image</text>
    </g>
  </defs>
  <image width='100' height='50' xlink:href='badlink.png'/>

Batik too shows some placeholder SVG content when an <image> has a bad
href, but I think it just falls back to a standalone SVG file rather
than an SVG fragment.  Being able to reference a fragment like above
would need some though regarding preserveAspectRatio, perhaps.

A less involved would be start off with an ‘image-fallback’ property
that takes values “none” and “auto” (not sure which the initial value
should be), where “auto” means the UA can draw whatever fallback
content it likes in the viewport established by the <image>, and “none”
would prevent that from being drawn.  We could then look at extending
that with a url() value later.


This would be different from the <image> fallback content approach we
discussed previously, but I think there would be use cases for both
approaches.

-- 
Cameron McCormack ≝ http://mcc.id.au/

Received on Monday, 23 March 2009 06:01:14 UTC