[uf-discuss] Re: rel=tag problem

[Ciaran McNulty]

On Mon, Mar 17, 2008 at 3:23 AM, David Meade <meade.dave at gmail.com> wrote:
>  But the specification clearly states that spaces should be encoded
>  with + or %20, and wordpress is encoding them as dashes.
>  http://microformats.org/wiki/rel-tag#Encoding_issues

It's a SHOULD, not a MUST - + or %20 is best practice,  but wordpress
is free to use a different scheme if they already have a system in
place that rel-tag is being bolted on to.

It would be great if they could change it, but we have to be aware
that the concept of tagging both predates and is more widely known
than rel-tag.

>  > this is a non-issue because the meaning of tags is
>  >  contextually dependent on the tag space, so there's no possibility for
>  >  standardization around what tags mean, whether they're single- or
>  >  multi-word.
>
>  I'm not sure I follow what you mean here.

It pretty much means aggregation is hard.  Aside from issues like
encoding of spaces, which we've been discussing here, there's a whole
host of other issues.

As an example, a wiki about metal might have 'lead' as a tag, whereas
a site about dogs might have a 'lead' tag that's a synonym for
'leash'.

This is the reason that URL prefixes are included in the semantics of
tagging - different tagspaces will associate different meanings with
different tags, compound or otherwise.  Aggregation where URLs are
discarded is always going to be a lossy process at best and that
applies to character encoding too.

-Ciaran McNulty