Re: Does SVG 1.0 define this?

On Wed, 14 Jul 2004, Dean Jackson wrote:
>>
>> If a UA finds XLink attributes on an SVG-namespaced element and those
>> attributes are not allowed per the SVG DTD, should the UA consider the
>> element to be in error or not? If not, should the UA just ignore the
>> attributes, or should it attempt to apply XLink semantics to the element?
>
> You propose three choices:
>
> 1. element in error
> 2. ignore
> 3. apply xlink
>
> I'm against 2. I'm not sure which of 3 or 1 is better, but both have
> problems. Especially on Mobile which won't have an xlink processor and
> usually is unable to test for incorrect unknown attributes (so can't do
> 1 either).
>
> What do you suggest?

I don't mind either way, so long as it is defined so I can write tests for
it. The best way would be to take the relevant tests in the SVG test suite
(assuming you have some -- if you don't make some) and then check what the
majority of UAs do, and codify that.

I highly doubt that most UAs actually implemented XLink. So I doubt 3 will
be common. Given how poorly the SVG error handling rules have been
implemented, I also doubt that they cause the document to abort
processing. So I doubt 1 will be common either. Thus I would guess that
the most common behaviour would be 2. This is a violation of XLink, of
course, but as you said, with hindsight XLink was probably not a good idea
for SVG.

The HTML/CSS error handling mindset is to ignore errors and pretend they
never happened, so if we're not to apply the XLink semantics, then I would
ignore the attributes.

But like I said. I don't mind either way, just so long as I know what
should happen so I can test it! :-)

Cheers,
-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'

Received on Wednesday, 14 July 2004 06:09:14 UTC