[uf-discuss] rel-tag links with #
kevinmarks at gmail.com
Fri Aug 3 13:12:45 PDT 2007
On Aug 3, 2007 12:23 PM, Taylor Cowan <taylor_cowan at yahoo.com> wrote:
> In using hAtom I was interested in populating the "@scheme" attribute of the Atom category.
> For example:
> It's interesting to me to know from where the tag comes, which community or folksonomy is it part of. It's valuable information that gets lost in the rel-tag parse as indicated by the spec today.
That is the tagspace part for the URL, eg for
htttp://technorati.com/tag/goat 'goat' is the tag and
'htttp://technorati.com/tag/' is the tagspace. This maps to the Atom
> Then I found that '#' anchors are ignored based on the defacto standards. (http://microformats.org/wiki/rel-tag-issues)
This is by design.
> What impressed by is that the defacto standards pretty much entail that the last token of the link (/a/b/lastoken) and the label ( ..>label</a> ) are nearly always the same. This holds for nearly every example I've looked at. The concept of having a label, a visual display for a tag, is completely foreign to my understanding of tags....and yet the rel-tag allows for this. That's so odd, because tags are normally just that...the tag and nothing but the tag. I'd also bring to the discussion the theme of visual human readable data...wherein it seems preferable to have the link, and the tag be the same:
> <a rel="tag" href="/mytags/weird">This is weird and rarely happens</a> (although perfectly acceptable by rel-tag)
> should be:
> <a rel="tag" href="/mytags/normal">normal</a> (this is the defacto standard, it's not weird like the previous example)
> and could also be
> <a rel="tag" href="/mytags#oktoo">oktoo</a> (still abides by URL standards)
no, this is tagging it 'mytags'
The Display text allows a human-readable tag where the underlying url
is less so eg <a
> What I find most ironic is that the defacto standards mentioned on the wiki, flickr and del.icio.us, don't even use rel-tag (or at least not yet). the HREF is valuable as a link to more information about the tag, from whence it came, related items, and the value of the <a> tag should contain the tag itself, as is the defacto standard and "ESTABLISHED PRACTICE" on both flickr and del.icio.us.
> I propose at the very least that '#' be allowed in the rel-tag spec...if we're writing weird parsers to snip of that last term based on '/', we might as well add '#' as a delimiter, and it would even be better if the spec can be reconciled with how tags are really used on the web, ie, the display text defines the tag.
The reason we chose path component was a) this was existing practice
and b) it provides some protection against using totally arbitrary
URLs as tagspaces, which creates spam jeopardy. You need to link to a
real tagspace that behaves sensibly if someone clicks on it.
More information about the microformats-discuss