-----Original Message-----
From: wikitech-l-bounces(a)lists.wikimedia.org
[mailto:wikitech-l-bounces@lists.wikimedia.org] On Behalf Of
Simetrical
Sent: 07 March 2008 20:23
To: Wikimedia developers
Subject: Re: [Wikitech-l] [Commons-l] Category
intersection:Newextensionavailable
On Fri, Mar 7, 2008 at 3:18 PM, Jared Williams
<jared.williams1(a)ntlworld.com> wrote:
Could be totally wrong in the database
that's why its
being changed.
It's still "right" technically. The results are accurately
reflecting the actual contents of the database at a definite
point in time.
That's all integrity asks for. Returning results that were
never correct just because of bad hashing algorithms is
something completely different. It's the difference between
your stock ticker being five minutes outdated, and generating
entirely random numbers sometimes.
To say there's no difference, or that the former makes the
latter okay, is ridiculous.
But having accuracy in a stock ticker, and wikipedia category intersection
are completely different and comparisons with between two are meaningless.
I think if giving an approximate answer is more scalable, then I think its
worth it.
A better comparison would be google, with all they're resources they still
approximate the number of seach results it expects
Jared