Should we exclude rel=nofollow by default?

[tunetheweb]

We've discovered that WooCommerce "add to basket" buttons are regular links. Now of course that breaks all sorts of GETs should be idempotent but such is the web.

They do implement a rel=nofollow to prevent search engines following such links. We're going to add it to our WordPress plugin, but I wonder if that's something we should do by default since it seems like a good sign that it's unexpected to be followed by bots and only by real humans.

Not sure if this is a spec question, or should be up to individual implementers so raising here first.

[tunetheweb]

Relevant WordPress issue: WordPress/performance#1140

[domenic]

I like this idea in theory. But I can't think of how we would reasonably specify it or explain it. "You said to prerender 'all links matching the /* URL pattern', but what we believe you meant was 'all links matching the /* URL pattern and without rel="nofollow"'. We'd need some concept of... default conditions which get anded onto any that are supplied?? And some way of undoing those???

Maybe it's best to just take this as good advice for when we create examples and documentation? Not sure.

[tunetheweb]

Well I was thinking it would only be for document rules and wouldn't be overridden. In the same way that "href_matches": "*" doesn't match external links for document rules even thought you said * (i.e. everything).

But we could add a "include-nofollow": true setting that defaults to false for those that did want to override it?

[domenic]

In the same way that "href_matches": "*" doesn't match external links for document rules even thought you said * (i.e. everything).

It does match :)

[tunetheweb]

It does match :)

I'm confused.

This example: https://prerender-demos.glitch.me/document-rules.html has this:

  <p><a href="next.html">This link should be eligible to be prerendered</a></p>
  <p><a href="next2.html">This link should also be eligible to be prerendered</a></p>
  <p><a href="https://www.example.com">This link is external and should not be eligible</a></p>

  <script type="speculationrules">
    {
      "prerender": [
        {
          "where": {"href_matches":"*"},
          "eagerness": "moderate"
        }
      ]
    }
  </script>

But shows this in devtools:

So the external link is excluded from the list of potential speculations? And that makes sense to me as the URL pattern has a base of the document AFAIK.

[domenic]

You're right, that's because * doesn't mean everything, it means URLs which match the current URL plus a * segment at the end.