On 02/01/2008, Stefan de Konink <[EMAIL PROTECTED]> wrote: > Tom Hughes schreef: > > On 02/01/2008, Stefan de Konink <[EMAIL PROTECTED]> wrote: > > > >> I wonder if it could be accepted that every day an automatic spell check > >> script would operate on the database and checks for less common 'name' > >> tags for automatic spelling corrections. > >> > >> select name, count(name) AS count from way_tags group by name order by > >> count, name; > > > > That query should only lock the API for an hour or so at a guess. > > In my setup: > > 89696.395 msec resulting in 10434 rows
...this would be a setup that has a name field in way_tags? unlike the actual live setup which doesn't? Even a 90 second lockup (and on the live system with other users it might be longer) is not really acceptable though, and if you use way_tags then it will be a lockup as that is a MyISAM table. I just did this on the live system: select k, count(k) AS count from current_way_tags group by k order by count, k which I assume is what you meant, and after two minutes I killed it because it had completely locked the API. > Yes, it isn't MySQL, don't worry, the machine has 8GB of memory and only > costed 390 euro. Would it be better if I provide a list of suggestions > and that this is applied to the database via the API or something faster? Huh? I thought that was a query you wanted to run on the live database? Which is a MySQL database... Tom (who is now very confused about what you are proposing) -- Tom Hughes ([EMAIL PROTECTED]) http://www.compton.nu/ _______________________________________________ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/cgi-bin/mailman/listinfo/dev
