Hacker News Reader

Comment by alextgordon | Hacker News Reader

I think the best way to do it would be an XML file referenced from a meta tag. This is a property of websites, not web pages, so it makes sense to put all the information in one place.

    <!-- page.html -->
    <meta name="msapplication-jumplist" content="http://arstechnica.com/jumplist.xml">
    
    <!-- jumplist.xml -->
    <?xml version="1.0" encoding="UTF-8"?>
    <jumplist>
        <name>Ars Technica</name>
        <starturl>http://arstechnica.com/</starturl>
        <tooltip>Ars Technica: Serving the technologist for 1.2 decades</tooltip>

        <task>
            <name>News</name>
            <action-uri>http://arstechnica.com/</action-uri>
            <icon-uri>http://arstechnica.com/favicon.ico</icon-uri>
        </task>
        <task>
            <name>Features</name>
            <action-uri>http://arstechnica.com/features/</action-uri>
            <icon-uri>http://static.arstechnica.net/ie-jump-menu/jump-features.ico</icon-uri>
        </task>
        <!-- etc -->
    </jumplist>

View on HN

Yeah, a reference to an external file would be a better way to do it. And HTML5 already has the mechanism for this sort of thing: the <link> element[1] and the "rel extensions" registry.

[1] http://www.whatwg.org/specs/web-apps/current-work/multipage/...

[2] http://wiki.whatwg.org/wiki/RelExtensions

Uh, doesn't that just add an extra round trip and yet another mandatory HTTP request? No thanks, I think we've enough of those.

> Uh, doesn't that just add an extra round trip and yet another mandatory HTTP request?

No. The request only happens if you decide to "pin" the site, or otherwise give it a first-class status on your machine (one could imagine the same by pining it to the OSX dock or making a site into a "standalone app", I think there was a browser doing that a while ago)

Even worse. That gives first-class status to the "HTML version" which contains links to all of the other "versions" of a webapp.

> That gives first-class status to the "HTML version"

Erm... you make absolutely no sense here?

Of course the "HTML version" (aka hello we're talking about web pages here) is the main resource, what else would be, unicorn and ponies?

Well, to be fair, HTML5 also already has the <meta> extension mechanism. I know you're not making the argument that one is more "standardsy" than the other; you're just pointing out that both approaches are equally "standard".

This also has the advantage of not bloating every HTML page with the jump list data. Using an external resource that is downloaded once at the time of pinning is clearly a technically superior mechanism and I'm surprised that it's not implemented like that already.

Oh, you mean like Aurora was going to use? That's Netscape Aurora, in 1998 http://xml.coverpages.org/revengeXML.html

The format was RDF, any any day now it's going to take over the web with all this nicely formed, semantic data!!

Though I'll never use IE as a daily browser, I do love the possibilities with this new SSB (Site Specific Browser) and jump list functionality. That said, I like your way of doing it better. It's much cleaner than the "spam the main file" method they're using now.

Shit, I thought JSON had taken over by now. Well, I guess html is XML, and you know they say XML is like violence; if it didn't solve your problem is because you didn't use enough...

Reply on news.ycombinator.com