[uf-new] OpenShapes - a tentative proposal for a Microformat for diagrams

[Alan Slater]

Hi there,

Thanks for the comments.

The work described was actually fairly well underway when I actuall 
"got" the Microformat approach of overloading base XHTML with "semantic" 
information - comparing the custom schema that I had been working on 
with a (directly equivalent) XHTML representation the latter seemed, to 
me at least, intuitively more appealing. However, I appreciate that the 
benefits aren't fantastically well thought out - which is why I marked 
the proposal as tentative.

I guess I'm particularly interested in being able to capture 
semantically rich diagrams (notably UML) within a "lowercase semantic 
web" approach.
> Your paper gives three reasons:
>
> 1. "The ability to embed diagrams into XHTML pages and retain validity
> without employing new namespaces etc." What's wrong with OBJECT and 
> IMG? If you want to inline diagram markup inside the same document, 
> you can retain XML validity by mixing XHTML with other XML languages 
> such as SVG. Why is avoiding namespaces such a crucial goal?
Well, I wouldn't say avoiding namespaces is "crucial" but I do find the 
simplicitly of microformats a rather welcome "breath of fresh" air after 
years of dealing with complex industry-specific XML "standards". YMMV

Part of floating this idea was to see whether there was any mileage in 
using an XHTML-based microformat - in terms of actual coding using a 
native schema is probably a bit easier!

Incidentally, although I don't mention this anwhere in the paper I will 
admit that this approach was probably heavily influenced by using pic 
and troff - but that's probably me showing my age! e.g.

http://www.troff.org/prog.html#pic
> 2. "Relative friendliness for search engine crawlers". Not as friendly 
> as a HTML long description of the same diagram, though.
Yes, but in the kind of collaborative environment that I see this 
potentially being useful (mainly graphical wikis), nobody is going to 
bother writing a separate "long description".
> 3. "The ability to render something if the content cannot be displayed
> graphically. Particularly important for accessibility – this should
> always be a major concern." Well, yes, but can you elaborate on how 
> such markup like your example is going to be meaningful to users of 
> existing user agents and assistive technologies, seeing as your links 
> have no link text and AFAIK no current consuming agent will do 
> anything special with rel="from" and rel="to"?
I think I copied the "links without link text" from one of the other 
Microformat examples... but I agree that this is no excuse!

Thanks again for the feedback.

Cheers

Alan