Question

我在查询中遇到问题，我在其中选择给定的记录时间跨度。

我选择的列是DATE类型。我已经绘制了这个列作为DateTime属性，查询有效，但速度很慢。

生成的查询如下所示:(由NHProfiler提供）

select kifkalende0_.KALENDER_MEDARBEJDER_ID as KALENDER1_119_0_,
   kifkalende1_.KALENDER_EMNE_ID        as KALENDER1_210_1_,
   kifkalende0_.OPDATERET_TIDSPUNKT     as OPDATERET2_119_0_,
   kifkalende0_.AENDRET                 as AENDRET119_0_,
   kifkalende0_.OPDATERET_AF            as OPDATERET4_119_0_,
   kifkalende0_.OPRETTET_AF             as OPRETTET5_119_0_,
   kifkalende0_.OPRETTET_TIDSPUNKT      as OPRETTET6_119_0_,
   kifkalende0_.SLETTET                 as SLETTET119_0_,
   kifkalende0_.KALENDER_EMNE_ID        as KALENDER8_119_0_,
   kifkalende0_.MEDARBEJDER_ID          as MEDARBEJ9_119_0_,
   kifkalende1_.OPDATERET_TIDSPUNKT     as OPDATERET2_210_1_,
   kifkalende1_.BESKRIVELSE             as BESKRIVE3_210_1_,
   kifkalende1_.DATO                    as DATO210_1_,
   kifkalende1_.ER_FRA_SAG              as ER5_210_1_,
   kifkalende1_.FRA_SAG_ID              as FRA6_210_1_,
   kifkalende1_.FRA_TABEL               as FRA7_210_1_,
   kifkalende1_.FRA_TID                 as FRA8_210_1_,
   kifkalende1_.OPDATERET_AF            as OPDATERET9_210_1_,
   kifkalende1_.OPRETTET_AF             as OPRETTET10_210_1_,
   kifkalende1_.OPRETTET_TIDSPUNKT      as OPRETTET11_210_1_,
   kifkalende1_.SAG_TYPE                as SAG12_210_1_,
   kifkalende1_.TIL_TID                 as TIL13_210_1_,
   kifkalende1_.YDERLIGERE_BESKRIVELSE  as YDERLIGERE14_210_1_,
   kifkalende1_.EMNE_ID                 as EMNE15_210_1_,
   kifkalende1_.PERSON_ID               as PERSON16_210_1_
from   "KIF_KALENDER_MEDARBEJDER" kifkalende0_
   left outer join "KIF_KALENDER_EMNE" kifkalende1_ on 
kifkalende0_.KALENDER_EMNE_ID = kifkalende1_.KALENDER_EMNE_ID,
   "KIF_KALENDER_EMNE" kifkalende2_
where  kifkalende0_.KALENDER_EMNE_ID = kifkalende2_.KALENDER_EMNE_ID
   and (kifkalende0_.MEDARBEJDER_ID in (7624 /* :p3 */,6226 
/* :p4 */,7382 /* :p5 */,5774 /* :p6 */, 5775 /* :p7 */,8259 
/* :p8 */,8218 /* :p9 */,9899 /* :p10 */, 6000 /* :p11 */,5779 
/* :p12 */,5780 /* :p13 */,5782 /* :p14 */, 5783 /* :p15 */,5784 
/* :p16 */,5785 /* :p17 */,5788 /* :p18 */, 5789 /* :p19 */,5790 
/* :p20 */,7341 /* :p21 */,8963 /* :p22 */, 10201 /* :p23 */,10388 
/* :p24 */))       
and kifkalende2_.DATO >= TIMESTAMP '2010-11-10 00:00:00.00' /* :p0 */
and kifkalende2_.DATO <= TIMESTAMP '2010-11-10 23:59:59.00' /* :p1 */
and (kifkalende0_.SLETTET = TIMESTAMP '1899-12-31 00:00:00.00' /* :p2 */
    or kifkalende0_.SLETTET is null);

在我们的数据库中，执行需要大约1500毫秒。

如果我们手动将查询更改为：

select kifkalende0_.KALENDER_MEDARBEJDER_ID as KALENDER1_119_0_,
   kifkalende1_.KALENDER_EMNE_ID        as KALENDER1_210_1_,
   kifkalende0_.OPDATERET_TIDSPUNKT     as OPDATERET2_119_0_,
   kifkalende0_.AENDRET                 as AENDRET119_0_,
   kifkalende0_.OPDATERET_AF            as OPDATERET4_119_0_,
   kifkalende0_.OPRETTET_AF             as OPRETTET5_119_0_,
   kifkalende0_.OPRETTET_TIDSPUNKT      as OPRETTET6_119_0_,
   kifkalende0_.SLETTET                 as SLETTET119_0_,
   kifkalende0_.KALENDER_EMNE_ID        as KALENDER8_119_0_,
   kifkalende0_.MEDARBEJDER_ID          as MEDARBEJ9_119_0_,
   kifkalende1_.OPDATERET_TIDSPUNKT     as OPDATERET2_210_1_,
   kifkalende1_.BESKRIVELSE             as BESKRIVE3_210_1_,
   kifkalende1_.DATO                    as DATO210_1_,
   kifkalende1_.ER_FRA_SAG              as ER5_210_1_,
   kifkalende1_.FRA_SAG_ID              as FRA6_210_1_,
   kifkalende1_.FRA_TABEL               as FRA7_210_1_,
   kifkalende1_.FRA_TID                 as FRA8_210_1_,
   kifkalende1_.OPDATERET_AF            as OPDATERET9_210_1_,
   kifkalende1_.OPRETTET_AF             as OPRETTET10_210_1_,
   kifkalende1_.OPRETTET_TIDSPUNKT      as OPRETTET11_210_1_,
   kifkalende1_.SAG_TYPE                as SAG12_210_1_,
   kifkalende1_.TIL_TID                 as TIL13_210_1_,
   kifkalende1_.YDERLIGERE_BESKRIVELSE  as YDERLIGERE14_210_1_,
   kifkalende1_.EMNE_ID                 as EMNE15_210_1_,
   kifkalende1_.PERSON_ID               as PERSON16_210_1_
from   "KIF_KALENDER_MEDARBEJDER" kifkalende0_
   left outer join "KIF_KALENDER_EMNE" kifkalende1_ on 
kifkalende0_.KALENDER_EMNE_ID = kifkalende1_.KALENDER_EMNE_ID,
   "KIF_KALENDER_EMNE" kifkalende2_
where  kifkalende0_.KALENDER_EMNE_ID = kifkalende2_.KALENDER_EMNE_ID
   and (kifkalende0_.MEDARBEJDER_ID in (7624 /* :p3 */,6226 
/* :p4 */,7382 /* :p5 */,5774 /* :p6 */, 5775 /* :p7 */,8259 
/* :p8 */,8218 /* :p9 */,9899 /* :p10 */, 6000 /* :p11 */,5779 
/* :p12 */,5780 /* :p13 */,5782 /* :p14 */, 5783 /* :p15 */,5784 
/* :p16 */,5785 /* :p17 */,5788 /* :p18 */, 5789 /* :p19 */,5790 
/* :p20 */,7341 /* :p21 */,8963 /* :p22 */, 10201 /* :p23 */,10388 
/* :p24 */))  
   and kifkalende2_.DATO>=to_date('10-11-2010 00:00:00', 'DD-MM-YYYY HH24:MI:SS')
   and kifkalende2_.DATO<=to_date('10-11-2010 23:59:59', 'DD-MM-YYYY HH24:MI:SS')
   and (kifkalende0_.SLETTET=to_date('31-12-1899 00:00:00', 'DD-MM-YYYY HH24:MI:SS')
   or kifkalende0_.SLETTET is null);

它在大约50ms内执行。

有没有办法让NHibernate生成to_date而不是日期比较的时间戳??

我对RegisterDateTimeTypeMappings的工作原理感到有点困惑 Oracle10gDialect，但我尝试扩展它，将方法改为

            protected override void RegisterDateTimeTypeMappings()
            {
                    RegisterColumnType(DbType.Date, "DATE");
                    //RegisterColumnType(DbType.DateTime, "TIMESTAMP(4)");
                    RegisterColumnType(DbType.DateTime, "DATE");
                    RegisterColumnType(DbType.Time, "TIMESTAMP(4)");
            }

但这没有帮助。

我们的环境是：

.net（C＃）4.0
NHibernate 3.1.0，主要通过Linq使用
针对Oracle 11g的ODP.Net 11.2.2.0

有人有任何建议吗？

谢谢， ./Daniel

Answer 1

我通过覆盖以下NHibernate类解决了我的问题。 NHProfiler仍将显示TIMESTAMP值，但基础参数将为DATE类型 - 如果您的oracle列也是DateTime，则该参数将是可索引的。将鼠标悬停在NHProfiler中的参数上以查看我的意思 - 类似于：：p0：TIMESTAMP - '2011-07-01 00：00：00.00'日期代替：p0：TIMESTAMP - '2011-07-01 00：00：00.00'时间戳。

我也在使用Oracle 11g，C＃.Net 4，Nhibernate 3.2。

public class OracleDataClientDriver2 : OracleDataClientDriver
{
    protected override void InitializeParameter(IDbDataParameter dbParam, string name, SqlType sqlType)
    {
        switch (sqlType.DbType)
        {
                //Timestamp columns not indexed by Oracle 11g date columns. - Use Date 
            case DbType.DateTime:
                base.InitializeParameter(dbParam, name, SqlTypeFactory.Date);
                break;
            default:
                base.InitializeParameter(dbParam, name, sqlType);
                break;
        }
    }
}

Answer 2

如果参数的类型正确，则根本不应进行任何转换。只有在字符串类型中包含时间戳时才需要TIMESTAMP或to_date（）。如果参数输入正确，则应如下所示：

select
       tab.COL as col
   from
       TABLE tab
   where
       tab.COL = :p0;
:p0 = 01.01.2000 00:00:00 [Type: DateTime (0)]

因此，请确保将DateTime个对象传递给查询。

NHibernate和DateTime映射的问题

2 个答案: