Re: microdata use cases and Getting data out of poorly written Web pages

Geoffrey Sneddon wrote:
> On 9 May 2009, at 14:59, Shelley Powers wrote:
> 
>> It would have been foolish to break down what requirements there are 
>> for vector graphics, then create use cases and proposals for every 
>> aspect (must support gradients, support for transforms, etc.), and 
>> then come up with some kludged together vector graphic support that 
>> isn't currently supported by any company, only because SVG is an 
>> "implementation detail".
> 
> We did actually do that, for what it's worth.

It might be helpful to provide evidence of that.

E.g. http://wiki.whatwg.org/wiki/New_Vocabularies lists some 
technology-independent use cases for that discussion, like "Writing a 
document by hand, with inline diagrams imported from a graphics package."

http://www.w3.org/mid/Pine.LNX.4.62.0803291940170.12640@hixie.dreamhostps.com 
then says:

> I haven't looked in detail at the vector graphics side of the problem yet. 
> Off-hand it looks like SVG would be a reasonably good fit for some of the 
> requirements, in particular ease of implementation and a reasonably 
> well-defined scripting environment, but it also isn't a perfect fit for 
> everything, for example it probably isn't supported in as wide a 
> collection of editors as, say, the Windows Metafile format, and isn't 
> supported in as wide an installed base of browsers as, say, VML. I do not 
> wish to prematurely limit our options, doing so would not be fair to the 
> Web authoring community.

http://www.w3.org/mid/Pine.LNX.4.62.0804060112390.18949@hixie.dreamhostps.com 
(proposed solutions to the use cases) later says:

> After briefly considering various formats, I concluded that the only
> reasonable choice for a vector graphics format in text/html was SVG. It is
> already supported to some extent by most major browsers, and it has
> reasonably well-defined DOM semantics.

For some other uses cases (e.g. "Writing a document by hand, with vector 
graphic icons."), the proposed solution was to not change the spec at 
all (e.g. because you can already solve it by putting SVG in <img>).

So it wasn't a case of considering use cases like "I want to draw 
something with circles" and "I want to draw something with gradients" 
and building up a new vector markup language. But it did start with use 
cases that could be solved without SVG, because there might be better 
solutions that don't involve SVG at all, and it was concluded that the 
benefits/costs of reusing SVG for at least one of the cases were clearly 
better than the benefits/costs of other possible solutions, and nobody 
argued against that conclusion, so it went in the spec.

-- 
Philip Taylor
pjt47@cam.ac.uk

Received on Saturday, 9 May 2009 15:28:21 UTC