Re: [eclipselink-users] Legacy database ddl generation with duplicate co

[yenan zhou <his.stone@xxxxxxxxx>]

ExercisePK is as simple:

[code]

@Embeddable
public class ExercisePK implements Serializable {
    @Basic(optional = false)
    @Column(name = "FinfoID")
    private String exerciseID;
    @Basic(optional = false)
    @Column(name = "FcorpId")
    private String corpID;

    public ExercisePK() {
    }

    public ExercisePK(String exerciseID, String corpID) {
        this.exerciseID = exerciseID;
        this.corpID = corpID;
    }

  ........ getters/setters/hash/equals....

}
[/code]

In the logging I see following line:

Internal Exception: java.sql.SQLException: Column already exists:
FCORPID in statement [CREATE TABLE MB_TMBGymInfo (FleaveDt
VARCHAR(255), Fremark VARCHAR(255), FCardID VARCHAR(255), FState
VARCHAR(255), FCardCorpID VARCHAR(255), FDoPlace VARCHAR(255), FComeDt
TIMESTAMP, FtypeName VARCHAR(255), Fkind VARCHAR(255), FRoomName
VARCHAR(255), FCardCorpName VARCHAR(255), FMemberName VARCHAR(255),
FMemberID VARCHAR(255), FinfoID VARCHAR(255) NOT NULL, FcorpId
VARCHAR(255) NOT NULL, FcorpID VARCHAR(255)]
Error Code: -27

While the line is a bit long, you'll see there're two FcorpId column
definition at the end of the create statement.

I think this is due to there's one def in ExercisePK and one in
@ManyToOne mapping in Exercise. So at the creation statge, both defs
generate a column definition in sql create statement. While if I set
insertable=true in @ManyToOne mapping, I'll also see two defs in sql
update statement too, and it leads to similar error but for updates.

For reads, it's not a problem as both defs will get there needed value
'coz I think select does not prevent duplicate projection attributes.

2010/2/24 Christopher Delahunt <christopher.delahunt@xxxxxxxxxx>:
> Hello,
>
> Seems odd.  Can you provide the ExercisePK class and the DDL statement that gives the problem (or the exception stack if it doesn't generate one)?
>
> Best Regards,
> Chris
>
> ----- Original Message -----
> From: his.stone@xxxxxxxxx
> To: eclipselink-users@xxxxxxxxxxx
> Sent: Wednesday, February 24, 2010 2:00:38 AM GMT -05:00 US/Canada Eastern
> Subject: Re: [eclipselink-users] Legacy database ddl generation with duplicate column definition.
>
>
> Well, you do have quick eyes. However, this is not the problem. This legacy
> database use "FcorpID" in exercise table and use "FCoprID" in member table.
> The design is such, hence the difference in mapping column names. It's not
> the error.
>
> I searched hibernate forum for similar questions, it seemed that hibernate
> has this issue too. I don't know if it's solved yet though.
>
> If I leave insertable and updatble to default, I can even not able to update
> table data since there're duplicate mappings. As I turn off these settings,
> updaing is working fine, but there lacks a way to notify also schema
> creation that only one mapping of the two is needed.
>
>
> christopher delahunt wrote:
>>
>> Hello,
>>
>> EclipseLink is case sensitive when it comes to processing column names,
>> but most databases are not.  If you have "FcorpID" in the join column as
>> the foreign key but "FCorpID" in the ExcercisePK mapping EclipseLink
>> will treat them as two separate fields instead of the same field.  This
>> could be why you are getting the exception on DDL generation, as
>> EclipseLink will try to create both fields separately.  Solution is to
>> make the field names the same in all definitions.
>>
>> Best Regards,
>> Chris
>>
>> zaya wrote:
>>> It seems to be a bug of eclipselink. I'm working on a legacy database.
>>> The
>>> mapping of table is like:
>>>
>>> [code]
>>> ... table definition
>>>
>>> @EmbeddedId
>>>     protected ExercisePK key;
>>>
>>> ... many other columns
>>>
>>> @ManyToOne
>>>     @JoinColumns({
>>>
>>> @JoinColumn(name="FMemberID",referencedColumnName="FMemberID",insertable=false,updatable=false),
>>>   @JoinColumn(name="FcorpID", referencedColumnName="FCorpID",
>>> insertable=false, updatable=false)
>>>     })
>>>     private AppMember appMember;
>>>
>>>
>>> [/code]
>>>
>>> The problem here is that the composite primary key is composed of the
>>> table
>>> auto id and also the "FcorpID" column. Because FcorpID is a global
>>> identifier which exits in every table to distingush similar data from
>>> other
>>> corporations.
>>>
>>> The mapping itself works fine. But when I want to test database using in
>>> memory db and generate ddl, eclipselink throw exception that it said the
>>> column definition "FcorpID" is duplicated and table is not created.
>>>
>>> I'm wondering if this is a bug or there's some configuration I can use to
>>> let eclipselink know that only one column here is needed like when
>>> persisting I turn off insertable and updatable to disable the second
>>> column's modification. Or is this a missing feature?
>>>
>> _______________________________________________
>> eclipselink-users mailing list
>> eclipselink-users@xxxxxxxxxxx
>> https://dev.eclipse.org/mailman/listinfo/eclipselink-users
>>
>>
>
> --
> View this message in context: http://old.nabble.com/Legacy-database-ddl-generation-with-duplicate-column-definition.-tp27662400p27714116.html
> Sent from the EclipseLink - Users mailing list archive at Nabble.com.
>
> _______________________________________________
> eclipselink-users mailing list
> eclipselink-users@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/eclipselink-users
> _______________________________________________
> eclipselink-users mailing list
> eclipselink-users@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/eclipselink-users
>