Really? I'd be curious to see how someone implements this. From my admittedly limited knowledge of the space, I see three different ways to tackling this. First, build out your own reservation system that uses your custom software to link the reservation system to the website and provide a vertical stack. Second, build plugins/hooks for existing electronic reservation/seating systems currently in use. Or third, work with the OpenTable's API.
OpenTable doesn't really have an official API. Urbanspoon has been touting Rez as an alternative but same story there. Will definitely let you know when we're ready for primetime.
Sorry to reply twice, but my previous reply was made quickly and I incorrectly recalled the wrong context. (I also recently mentioned using an extension to view PDF's in Google's Document Viewer.)
The extension I was referring to in my grandparent comment is:
(Note that there is another, older "Re-Pagination" extension that was abandoned. IIRC, this one picked it up and updated it.)
PLEASE BE CAREFUL if you use this extension. If you do not choose a limited page count, it is easy to unduly load or overload a server. This include the HN server, and such request behavior will get you cut off or banned (rightly so; please work to preserve HN resources).
I use it, infrequently, a time or two a day to load the first few top pages of results, which I then work through at my leisure.
I also use a bookmark to transform all the link href values to open in new pages/tabs. That way, I don't lose the browser-generated multiple-page page (as it were) by forgetting to right click or shift-click to open a link that appears on it in a new tab.
At the time I adopted it, IIRC I read through the code enough to determine it was not malicious. I've been perhaps lax in not checking all of the updates.
I see its max version is listed as 3.6 . I'm actually still on 3.6; when I go to 5, I guess I'll see whether it still works with max version overridden.
Web hooks would be nice. Open-sourcing would be GREAT! Especially for those of us who prefer to host their own da(y)ta.
Just downloaded it, and it is something I've been wanting to build for a long time, but just didn't get around to it. I would gladly collaborate, would you decide to open-source it.
In practice though, is the degree of connectedness really ever that high? I was under the impression that many real world phenomena (from social networks to biological interaction networks) exhibit small-world phenomena and you tend to get more well-connected subgraphs that are connected by important hub nodes. So it would seem to me like applying some careful heuristics (and keeping track of important bottlenecks/hubs) would still allow you to do computations in chunks.
It's not, a lot of the time, but in order to take advantage of that you generally need some domain-specific hack. More general solutions generally assume the possibility of interconnectedness at any point, ie, you can't rule it out, which means you'll need some kind of fast internode lookup mechanism as mentioned by pjscott.
