Apr 5, 2004
Hi,
Recently, I found a bug in Microsoft's official SQL Server2000 JDBC driver.
It seemed to lie in the Statement's setDate() method. Maybe they simply assumed
all date are represented in 12-hours format.
The test code is below:
public void testMisc() throws Exception {
Connection conn = TransactionManager.getConnection();
String sql = "select count(salescheck0_.ID) as x0_0_ " +
"from T_CHK_OUT_BILL_TEST salescheck0_ " +
"where (salescheck0_.EX_TIME>?)" +
"and(salescheck0_.EX_TIME<?)";
PreparedStatement stmt = conn.prepareCall(sql);
SimpleDateFormat sdf = new SimpleDateFormat("yyyy-MM-dd HH:mm:ss");
Date timefrom = sdf.parse("2004-04-02 00:00:00.000");
Date timeto = sdf.parse("2004-04-02 23:00:00.000");
stmt.setDate(1, new java.sql.Date(timefrom.getTime()));
stmt.setDate(2, new java.sql.Date(timeto.getTime()));
// stmt.setString(1, "2004-04-02 00:00:00.000" );
// stmt.setString(2, "2004-04-02 23:59:59.000" );
ResultSet rs = stmt.executeQuery();
if(rs.next()) {
int count = rs.getInt(1);
System.out.println("count = " + count);
}
}
The test data is following:
ID, SO_ID,WH_ID,CLIENT_ID,CODE,OPERATOR,CHK_OUT_TIME, STATUS,CREATE_TIME,DELIVER_TYPE
86179,67521,1,661,SLCCK200404020016,xuwei,2004-4-2 12:15,20,2004-4-2 9:40,10
86229,67566,1,118,SLCCK200404020066,xuwei,2004-4-2 12:23,20,2004-4-2 9:41,10
86231,67568,1,130,SLCCK200404020068,xuwei,2004-4-2 12:23,20,2004-4-2 9:41,10
86232,67569,1,134,SLCCK200404020069,xuyili,2004-4-2 11:08,20,2004-4-2 9:41,10
86233,67571,1,151,SLCCK200404020070,xuwei,2004-4-2 12:23,20,2004-4-2 9:41,10
86234,67572,1,55,SLCCK200404020071,xuwei,2004-4-2 12:23,20,2004-4-2 9:41,10
86235,67573,1,71,SLCCK200404020072,xuwei,2004-4-2 12:23,20,2004-4-2 9:41,10
86252,67590,1,298,SLCCK200404020089,xuwei,2004-4-2 12:33,20,2004-4-2 9:41,10
86268,67576,1,25,SLCCK200404020105,xuwei,2004-4-2 12:23,20,2004-4-2 10:18,10
86269,67588,1,271,SLCCK200404020106,xuwei,2004-4-2 12:33,20,2004-4-2 10:20,10
86270,67506,1,42,SLCCK200404020107,xuwei,2004-4-2 12:15,20,2004-4-2 10:21,10
86271,67531,1,751,SLCCK200404020108,xuyili,2004-4-2 11:28,20,2004-4-2 10:22,10
86272,67570,1,149,SLCCK200404020109,xuwei,2004-4-2 12:23,20,2004-4-2 10:29,10
86273,67600,1,6,SLCCK200404020110,xuwei,2004-4-2 12:15,20,2004-4-2 10:40,10
86373,67665,3,686,SLCCK200404020210,root,2004-4-2 17:18,20,2004-4-2 16:32,10
86374,67657,3,883,SLCCK200404020211,root,2004-4-2 17:18,20,2004-4-2 16:32,10
86375,67666,3,686,SLCCK200404020212,root,2004-4-2 17:18,20,2004-4-2 16:32,10
86376,54799,1,395,SLCCK200404020213,zhouliyi,2004-4-2 19:52,20,2004-4-2 19:51,10
86377,55383,1,341,SLCCK200404020214,zhoulingyi,200 4-4-2 19:52,20,2004-4-2 19:51,10
Given the code and data, the expected output ought to be "count = 19"
But it's "count = 19" instead.
However, if you set date by using alternative method like the commented statements,
you can get the correct result.
Does anybody know if there exists any patch?
Could we expect Mircosoft to fix this bug? If so, where could I submit this bug report?
Regards,
Justin
View 3 Replies
View Related
Jul 23, 2005
Hello JoeOver the past several years, I have found your responses to jdbcusage/driver related issues to be extremely helpful. I am sure thatyou're very busy so I will make my question as brief as possible - thereason I didn't post this to a newsgroup is that I don't want to hearany more hype or links to benchmarks (since the vast majority of thebenchmarks don't test concurent/mixed-transaction environments). (I amalso posting a modified version of this message to some newsgroups, butbased on past responses - don't have too much hope that these will leadto a resolution.) If the only way to answer this question would be tohave a phone conversation, my company would be more than happy to pay aconsulting-service fee for your time.My situation:My company has a enterprise level web-app that targets SQL Server. Wedon't use J2EE - so we stick to a simple to administer web containers(JRun,Tomcat). Recently, we have been getting worse and worseperformance from our jdbc driver (we use the free MS jdbc driver) -things like strange transaction resource handling, chopy overallperformance, etc.. I have spent significant time tweaking it (followingvarious advice - many times yours - on newgroups... I can go intodetails, but I don't want to take up your time). As the project'sarchitect, I need to do something about this problem - but varioushigh-level tunnings that I have done to the way we use connections withthis driver haven't significantly improved perfomance under normaleveryday load. (Our queries often span tables with millions of records,and are relatively dynamic. We use a seperate pool of autocommit-offconnections for writes, and autocommit-on connecions for reads)So, then, my question - in your expert opinion - what is the BESTdriver from SQL Server 2000 given the following needs:Things I need it to have are solid jdbc 1.0, solid transactionhandling, decent concurrent load handling, stable implementation ofresource handling (i.e. auto closing result sets when parent statementsclose, closing of result sets & statements when connections close) andsupport for multiple open statements/result sets per connection.Things I do not have a direct need for:Connection Pooling (I keep my own pool of open connections), DataSourcesupport, distributed transactions or 2-phase commit support, RowSetvariants, jdbc 3.0 autokey generation, J2EE compliance. We do make useof updatable result sets - but I don't care if it isn't supported sinceI can still use MS driver for those, as they are mainly done it batchjobs).Things that are nice:It would also be good to have a driver that can handle prep statementcaching on a driver level (vs connection level) although this isn't anecessity, and if it is available - it would need to have aconfigurable caching strategy (or have a way to be turned off :)I very much appreciate your help in advanceThanksGary BogachkovSystem ArchitectStericycle Direct Return
View 1 Replies
View Related
