On Tue, Jan 27, 2009 at 11:29 AM, Steve Summit <s...@eskimo.com> wrote:
> Jeff Ferland wrote: > > You'll need a quite impressive machine to host even just the current > > revisions of the wiki. Expect to expend 10s to even hundreds of > > gigabytes on the database alone for Wikipedia using only the current > > versions. > > No, no, no. You're looking at it all wrong. That's the sucker's > way of doing it. > > If you're smart, you put up a simple page with a text box labeled > "Wikipedia search", and whenever someone types a query into > the box and submits it, you ship the query over to the Wikimedia > servers, and then slurp back the response, and display it back > to the original submitter. That way only Wikimedia has to worry > about all those pesky gigabyte-level database hosting requirements, > while you get all the glory. > > This appears to be what the questioner is asking about. > Let's AGF a bit... Even if someone with a not particularly Wikipedia goal in life links to one of our searches from their page, all the resultant search result links are back into Wikipedia. If people have a question about something, and want to look it up, does it really matter if they go to Wikipedia's front page and click "search" versus doing so in another context? We're providing an information resource - other sites can and often do link to our articles (quite appropriately). Why not link to our search? The search link should in fairness tell people what they're getting, sure, but that's more of a website-to-end-user disclosure problem than a problem for us. Google switching to use our search would crush us, obviously. As would AOL. But J. Random site? Seems like an ok thing, to me. -- -george william herbert george.herb...@gmail.com _______________________________________________ Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l