TableJDBCSeq for GeneratedValue.TABLE did not handle initialValue correctly

• Key: OPENJPA-284
• Type: Bug
• Status: Closed
• Resolution: Fixed
• Priority: Major
• Assignee: Unassigned
• Reporter: Teresa Kan
• Component/s: jdbc
• Affects Version/s: 1.0.0
• Fix Version/s: 1.0.0

Description

When I tested the GeneratedValue strategy on TABLE, and I found a bug in the TableJDBCSeq that did not set the InitalValue. Therefore, the sequence always start from 1.
Here are the bugs in the TableJDBCSeq:
1) InsertSequence method:
       SQLBuffer insert = new SQLBuffer(dict).append("INSERT INTO ").
            append(_pkColumn.getTable()).append(" (").
            append(_pkColumn).append(", ").append(_seqColumn).
            append(") VALUES (").
            appendValue(pk, _pkColumn).append(", ").
            appendValue(Numbers.valueOf(1), _seqColumn).append(")"); --> Always set the initial value to 1.
2) no getter and setter on the InitialValue.

Here is the annotation:
@TableGenerator(name="Dog_Gen", table ="ID_Gen", pkColumnName="GEN_NAME", valueColumnName="GEN_VAL",pkColumnValue="ID2",initialValue=20,allocationSize=10)
@GeneratedValue(strategy=GenerationType.TABLE, generator="Dog_Gen")
private int id2;

The initial value always started from 1 with the current openjpa implementation.
The fix will be like this:
1) add getter and setter of InitialValue in TableJDBCSeq .
2) add int _initValue variable.
3) Change the InsertSequence method:
       SQLBuffer insert = new SQLBuffer(dict).append("INSERT INTO ").
            append(_pkColumn.getTable()).append(" (").
            append(_pkColumn).append(", ").append(_seqColumn).
            append(") VALUES (").
            appendValue(pk, _pkColumn).append(", ").
            appendValue(_intValue, _seqColumn).append(")"); ---> change to use the initValue instead of 1.

Comments

[Teresa Kan]

The patch is provided for this jiar report