On 05-08-2010 16:38:12 +0200, Jan Rittinger wrote:
On Aug 5, 2010, at 15:48, Fabian Groffen wrote:
On 05-08-2010 15:20:07 +0200, Manuel Mayr wrote:
PS.: I'm not sure if it is really the size of the SQL buffer since the query works when I modify it to run it against sys.tables, but it doesn't seem to return any result. Even prefix the query with "plan" doesn't come back within minutes. Maybe a performance bottleneck in the compiler or planner?
Could this be bug #2595 by chance?
You are not authorized to access bug #2595.
Hmm, yeah, sorry. We cannot open up that one. I can post the relevant bits though: The details are in bug #2595. Using sql_debug=65536 we can now (temporarily) disable the expensive or-style optimisation of selection push-downs. http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=897717064f69 The query in question actually uses a huge number of args in an IN query (... IN (x1, x2, ..., xlarge)), which looks like yours.