[uf-discuss] hReview feedback

[Ryan King]

On Jan 11, 2006, at 10:12 AM, Mark Nottingham wrote:

> (I had to restrain myself from Subject: hReview Review)
>
> In general, I like hReview a lot, and it fits the use case I'm  
> working on nicely.
>
> However, it's currently too restrictive in a couple of places; in  
> particular, it requires the reviewer and the date reviewed to be  
> explicitly included in the content. While I understand that this  
> information is useful -- or even required -- for some uses of  
> hReview, it may not be for others.

You have a good point about the reviewer. We may want to consider  
extracting or defering the problem. Date is another issue. See below.

> For example, a review's author might be inferred from the Web site  
> it's hosted on. The date of the review could be inferred from the  
> HTTP Last-Modified header,

Actually, no. The semantics of dtreviewed are not the same as the  
last modified date of the document.

dtreviewed means 'when the review was done,' not 'when the document  
was written. For example, if you got to a restaraunt on 2006-01-10,  
but don't write and publish the review until 2006-01-20, then  
dtreviewed would still by 2006-01-10.

The motivation behind this is that reviews will sometimes only be  
relevant within a certain timeframe.

> it could be unknown, or it could be irrelevant (e.g., a review of a  
> recipe).

Despite my previous comment, I think you're right- there are classes  
of reviews for which date has less significance.

> In general, I don't think that formats that are to be used on the  
> open Web should have lots of mandatory fields; it raises the bar  
> for authoring them (as well as repurposing existing data to them),  
> and ends up restricting how people use them, because of the  
> unforeseen consequences of doing so. I think this is the biggest  
> mistake we made in Atom; rather than making it an open format, we  
> required a bunch of elements that really weren't always required.

Your general point here is good. A nice thing about http:// 
microformats.org/wiki/hcard, for example, is that we have only one  
required field, N.

> Anyway, keep up the good work, and please make these fields optional.

Let's discuss the implications.

Also, if you would, could you put a note on http://microformats.org/ 
wiki/hreview-issues about this, so that we can keep track of it?

Thanks,
ryan
--
Ryan King
ryan at technorati.com