URI profiles [was RE: [uf-discuss] Comments from IBM/Lotus rep about Microformats]

[Joe Andrieu]

Scott Reynen wrote:
> On Dec 12, 2006, at 1:38 PM, Joe Andrieu wrote:
> > As I understand it profile URIs are not required.
> >
> > If so, the parser cannot distinguish between wild semantic 
> HTML and an 
> > hCard.
> 
> Profile URIs are not required for publishers, but parsers are 
> free to  
> ignore HTML without profile URIs, and I think it's reasonable to  
> expect them to start doing that if name conflict becomes more than a  
> hypothetical problem.  This mirrors how natural language works.   
> Until there is some need for clarification, we assume everyone knows  
> what we mean.  Then there is a need for clarification, we clarify.   
> No one goes around defining every word before they use it, and I  
> don't think we can expect publishers to behave differently with HTML  
> symbols.  We could require profile URIs, but that won't make anyone  
> use them.  OI think only a practical need for disambiguation will do  
> that.

Making people use them is not the same as clarifying in a spec what
should be done, must be done, and what is optional.  If we are
specifying that parsers can ignore non-profiled semantic HTML that looks
like microformats, we are essentially saying parsers can ignore
non-profiled microformats, since you can't tell the difference. Which
means that URI profiles are /effectively/ required if you want to be
assured that standards-compliant parsers will pick them up your
microformats.

Yea!  I think profiles are great.  So, why not formalize the
requirement?

If authors write non-compliant code (without the profile), *GASP*--who
would ever do that?--then they will have reason to understand why the
parsers ignore it. Just like if they write non-compliant HTML, they can
understand why it doesn't work. (Ahem, we'll ignore the issue of
non-compliant browsers).

However, without the profile requirement, authors have no reason to
expect that parsers won't pick up their "standards-compliant"
microformats.

100% compliant code should work with 100% compliant parsers. That seems
self-evident. Does it make sense to allow compliant parsers to ignore
compliant microformats?

-j

--
Joe Andrieu
joe at andrieu.net
+1 (805) 705-8651