Directory traversal (../) and anchor text: bad combo?

Is it best to use absolute url's in your links, or limit traversal?

[mincklerstraat]

In the days people still muttered about alltheweb results, I noticed that many of my pages were listed not only as example.com/index.html, but also example.com/ and example.com - three listings for one page. Naturally one asks whether the spider in question isn't considering punishing you for identical content in such a scenario.

I'm not worried about alltheweb now, but am making a content managed site for a client that's multilingual, and there are a whole lot of '../../this.html' '../that.html' links, though I could probably rewrite a lot of the code to make these links less traversal-oriented. So the question is: has anyone seen differences in the serps between links like "/this.html" and "../../this.html", when it's important that the anchor text gets counted for those links? Or are most spiders worth anything smart enough to register traversed links back to their real url?

[kaled]

If I were implementing a search engine, I would normalise all link urls. Thus relative and absolute links would behave identically because they would be stored identically in the relevant parts of the database.

Kaled.

[killroy]

Since the server receives usually no reliable info about your current location, it cannot follow traversals. That means the browser is resolving them and sending back the real URLs. Which makes it a safe assumption that spiders do the same.

SN

[mincklerstraat]

Thx for the advice! Rather happy too about not having to rescript all the links.

[abates]

OTOH there are a few really buggy spiders out there which seem to have a problem with relative URLs :)