On Thu, Jan 30, 2003 at 12:37:34PM +0100, Erik Moeller wrote:
On Don, 2003-01-30 at 12:17, Jimmy Wales wrote:
Brion Vibber wrote:
Well, an update (counter increment) comes with
every page view, and we
often have several edits per minute.
What's the wikipedia-purpose of a counter increment on every page
view? Just to know which pages are popular? We could get the same
information easily enough from a log analysis program, which I could
offload to a less-busy server. Or we could just do without the
information easily enough. It's not really that useful. Or am I
missing something?
Even if this is thought by Jan and others to be only a small part of
the performance problem, it's probably philosophically good for us to
get in the habit of thinking "unless this feature is free or really
really useful, we should do without it". That is, we already have a
bad case of feature-itis, and taking a hardline on such things might
be very helpful.
Yeah, the counters are wildly inaccurate anyway because they've been
disabled so often. Let's just get rid of them entirely. A most popular
ranking by URL would still be neat though.
I couldn't agree more. Whatever type of locking we are going to use,
restricting ourselves to read-only operations for page views is better for
concurrency and performance.
-- Jan Hidders
.---------------------------------------------------------------------.
| Post-doctoral researcher e-mail: jan.hidders(a)ua.ac.be |
| Dept. Math. & Computer Science tel: (+32) 3 820 2427 |
| University of Antwerp, room J1.06 fax: (+32) 3 820 2421 |
| (UIA) Universiteitsplein 1, B-2610 Wilrijk - Antwerpen (Belgium) |
`---------------------------------------------------------------------'