Mailing List Archive

> It strikes me that some of our nice "extra" features, like Most
> Wanted, are slow and could also be just as useful (or nearly so) if
> they were not generated 'realtime', but rather if they were generated
> only once per day.

i like this idea

Jimmy Wales wrote:

>It strikes me that some of our nice "extra" features, like Most
>Wanted, are slow and could also be just as useful (or nearly so) if
>they were not generated 'realtime', but rather if they were generated
>only once per day.
>
>
"Most Wanted" is already cached when invoked, and the cache is used in
"miser mode". That works good for Most Wanted, as the order of articles
doesn't change very fast, and created articles are clearly visible.

This will work less well on other "extra" features like "Orphaned
Pages". If one de-orphaned a page, it will still show on the cached version.

Magnus

On Mon, Feb 03, 2003 at 05:51:23AM -0800, Jimmy Wales wrote:
> It strikes me that some of our nice "extra" features, like Most
> Wanted, are slow and could also be just as useful (or nearly so) if
> they were not generated 'realtime', but rather if they were generated
> only once per day.

There's really still a lot we can do in terms of database optimization. I
have just installed Wikipedia at home again, and still have access with my
Sourceforge account (I think), so if Brion wants me to I could actually
start making my hands dirty again. :-) But when messing with the database
good coordination is extra important, so it should be very clear who is
working on what.

-- Jan Hidders

> On Mon, Feb 03, 2003 at 05:51:23AM -0800, Jimmy Wales wrote:
>> It strikes me that some of our nice "extra" features, like Most
>> Wanted, are slow and could also be just as useful (or nearly so) if
>> they were not generated 'realtime', but rather if they were generated
>> only once per day.

> There's really still a lot we can do in terms of database optimization. I
> have just installed Wikipedia at home again, and still have access with my
> Sourceforge account (I think), so if Brion wants me to I could actually
> start making my hands dirty again. :-) But when messing with the database
> good coordination is extra important, so it should be very clear who is
> working on what.

If you have a good connection, I suggest you import both the CUR and OLD
tables of the English Wikipedia. Some of the queries are dog slow even
when locally run.
http://www.wikipedia.org/wiki/Wikipedia%3ADatabase_download

Regarding coordination, you mostly need to check with Brion. Us others
primarily work on creating new problems ;-)

Thanks for helping out,

Erik

On Tue, Feb 04, 2003 at 11:40:00PM +0100, Erik Moeller wrote:
>
> If you have a good connection, I suggest you import both the CUR and OLD
> tables of the English Wikipedia. Some of the queries are dog slow even
> when locally run.
> http://www.wikipedia.org/wiki/Wikipedia%3ADatabase_download

They are being downloaded as I type this, and yes, I have a fast connection.
:-) Thanks for the URL. Now let's hope my ISP doesn't think I'm downloading
DVDs. :-)

-- Jan Hidders