db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Rick Hillegas <rick.hille...@oracle.com>
Subject outstanding bugs in the automatic calculation of statistics
Date Wed, 09 Mar 2011 14:13:02 GMT
Here are the open bugs in this area which I know about. Are there others?

o DERBY-5045: The best theory about this problem is that there is a 
pre-existing DataDictionary bug which fails to scour cruft out of 
SYSSTATISTICS after a table is dropped. The consequence of this 
pre-existing bug is that harmless junk tuples may slowly accumulate in 
SYSSTATISTICS for applications which have not converted from using DROP 
TABLE to using TRUNCATE TABLE. This does not appear to be a regression 
or even a user-visible problem.

o DERBY-5097: This appears to be a timing-related problem in which 
automatically calculated statistics are not computed soon enough. There 
may be a real istat bug here or this may be a heisenbug in the test 
itself. In any event, this does not appear to be a regression. The 
consequence of this problem is that istat may not deliver its benefits 
fast enough in some situations. This will not cause wrong results and 
this will not cause queries to run any slower than they did in previous 
releases.

Thanks,
-Rick

Mime
View raw message