HSQLDB: INSERT INTO ... (SELECT NULL, * FROM) 导致 "duplicate column name"

Question

在 HSQLDB 中，我有两个 TEXT tables，映射到 CSV 文件。查看项目 CSV Cruncher。

CREATE TEXT TABLE session_telephony_pins ( Op VARCHAR(4092), id ... )

我正在尝试向此 "input" table 添加一列，以便结果对每个 table.

都有一个唯一的 ID

所以这看起来很合乎逻辑：添加定义为序列的列，

CREATE TEXT TABLE output ( crunchCounter BIGINT
  GENERATED BY DEFAULT AS SEQUENCE crunchCounter PRIMARY KEY, op VARCHAR(4092), ... )

然后使用null让DBselect的值为：

INSERT INTO output (SELECT NULL AS crunchCounter, * FROM apollo_session_occurrence)
-- I also tried withhout `AS ...`

但是如果失败：

SQLSyntaxErrorException duplicate column name in derived table

表和列类型：

* APOLLO_SESSION_OCCURRENCE
  - OP                           CHARACTER VARYING
  - SESSION_OCCURRENCE_ID        SMALLINT
  - SESSION_ID                   SMALLINT
  - START_TIME                   TIMESTAMP
  - END_TIME                     TIMESTAMP
  - UID                          UUID
* OUTPUT
  - CRUNCHCOUNTER                BIGINT
  - OP                           CHARACTER VARYING
  - SESSION_OCCURRENCE_ID        CHARACTER VARYING
  - SESSION_ID                   CHARACTER VARYING
  - START_TIME                   CHARACTER VARYING
  - END_TIME                     CHARACTER VARYING
  - UID                          CHARACTER VARYING

我认为在 MySQL 这会起作用，但尚未测试。

类型应该不是问题。如果我删除 crunchCounter 列并只执行 INSERT INTO output (SELECT * FROM apollo_session_occurrence)，它工作正常。

理论上我可以使用 NEXT VALUE FOR crunchCounter 但这不会更改列元数据。当 HSQLDB 检查元数据时会发生这种情况。

这是怎么回事？而且，

我应该如何用唯一列填充 output table？

编辑：有趣的是，它与其他 tables 和特别命名的结果列一起工作得很好：

INSERT INTO output (SELECT NULL AS crunchCounter,  jobName, buildNumber, config, ar, arFile, deployDur, warmupDur, scale,
  CAST(warmupDur AS DOUBLE) / CAST(deployDur AS DOUBLE) AS warmupSlower
  FROM concat ORDER BY deployDur)

* CONCAT
 - JOBNAME                      CHARACTER VARYING
 - BUILDNUMBER                  SMALLINT
 - CONFIG                       CHARACTER VARYING
 - AR                           CHARACTER VARYING
 - ARFILE                       CHARACTER VARYING
 - DEPLOYDUR                    SMALLINT
 - WARMUPDUR                    SMALLINT
 - SCALE                        SMALLINT
* OUTPUT
 - CRUNCHCOUNTER                BIGINT
 - JOBNAME                      CHARACTER VARYING
 - BUILDNUMBER                  CHARACTER VARYING
 - CONFIG                       CHARACTER VARYING
 - AR                           CHARACTER VARYING
 - ARFILE                       CHARACTER VARYING
 - DEPLOYDUR                    CHARACTER VARYING
 - WARMUPDUR                    CHARACTER VARYING
 - SCALE                        CHARACTER VARYING
 - WARMUPSLOWER                 CHARACTER VARYING

开始看起来像一个 HSQLDB 错误。

Answer 1

想通了。问题在于 HSQLDB 将 SELECT 1 AS foo, * FROM myTable 中的 * 扩展为 foo, foo, myTable.col1, ...。

这是 IMO 的一个错误。编辑：不是错误，但 HSQLDB 可以提供更好的错误消息。

解决方法：使用合格的通配符。

INSERT INTO output (SELECT NULL AS crunchCounter, 
        apollo_session_occurrence.* 
   FROM apollo_session_occurrence ...)

HSQLDB: INSERT INTO ... (SELECT NULL, * FROM) 导致 "duplicate column name"

HSQLDB: INSERT INTO ... (SELECT NULL, * FROM) leads to "duplicate column name"

sql

hsqldb