The Underscore Archives

> Not so sure there's *that* much going on.
>
> For example, the 'history recognition' can only pull a specific URI —
> you're not going to be able to tell if someone's been an *any* page
> on google.com (I just ran a Google search earlier this morning, and
> am on Yahoo! sites all day, but both are 'unvisited'). Rather, they
> indicate whether someone was on the specific page specified in the
> href attribute. One would have to put an awful lot of links into a
> page to cover even a significant subset of google.com or yahoo.com or
> whatever. And the attribute selectors are actually superfluous; you
> could just use custom classes on the links to achieve the same effect.
>
> Really, there's not so much going on here besides some clever use of
> the :visited pseudo-class, a single Gecko-only property and a few
> HTTP requests. With the exception of the Gecko-only property, most of
> this has been possible from Netscape 4.x forward.
>
> And the 'CSS Attacks' page <http://www.businessinfo.co.uk/labs/
> css_attacks/css_scripting.php> is utter rubbish. It's just using
> the :visited pseudo-class to set margin (or maybe border, padding,
> top, right, etc.) properties. People have been doing that for *years*
> (even before Eric Meyer's 'Complex Spiral' demos in 2001).
>
> Maybe I'm missing something, but I'm really not impressed. When he
> can get some sort of control statements working beyond limited if/
> else branching I'll be more impressed. Until then…meh.

And even then, capturing that a user has visited google.com or  
yahoo.com, there's still not a lot they can do with this data, except  
display the fact it in the browser window.