Row cannot be located for updating

Hi,

It is not short of time to use postgresql for me, but recently got a problem
about record update as the title says.

I have encountered the same error before and all been resolved, but this
time it is a bit specific, the Internet couldn't help me to confirm the
problem.

I record the environment and the process of validation as follows.

Hope to get help.

1.Enviroment:

.PostgreSQL:9.3

.PostgreSQL ANSI 8.04.02.00

.client_encoding: GBK

2.Test Table and Data:

Create t_test_001 and insert 4 records for test.

The script of create as follow:

CREATE TABLE t_test_001

(

id integer NOT NULL,

name character varying(4),

CONSTRAINT pk_test_001 PRIMARY KEY (id)

);

Insert data.

(1)insert into t_test_001(id,name) values(1,'大'); --- without trailling
blank char_length=1

(2)insert into t_test_001(id,name) values(2,'大 '); --- with 1 trailling
blank char_length=2

(3)insert into t_test_001(id,name) values(3,'大 '); --- with 2 trailling
blank char_length=3

(4)insert into t_test_001(id,name) values(4,'大 '); --- with 3 trailling
blank char_length=4

notice the length of column id is 4.

3.Testing

Using ADO Recordset API such as Open,PutCollet,Update to update a specific
record.

The following code uses an ado wrapped class, but can be corresponding to
the original method.

string sql = "select name from t_test_001 where id=1"; ///< id=1 locate
the record want to update,replace 1 with the one want to update.

string name = "a";

pRecordset = pdbor->Query(sql.c_str(),adOpenStatic,adLockOptimistic);

if (!pRecordset->PutCollect("name",name.c_str()))

return -1;

if (!pRecordset->Update())

return -1;

change id in code above,execute for update every record in sequence.

the results showed that it is ok for records with id=1,2,3,but the record
with id=4 raise error as follow:

无法为更新定位行。一些值可能已在最后一次读取后已更改.

(Row cannot be located for updating. some values may have been changed since
it was last read.)

i found serveral factors to bring this error:

.varchar column contains Chinese char

.with trailling blank

.length of data is width of the column

how to explain and resolve it?

4.Additions

.when i replace with PostgreSQL Unicode Driver,it's ok.

.i would not set Update Criteria to adCriteriaKey to work around
this,because it's not sure that all query always contains a primary key
column.

Thank you,

whereh