Re: Teradata using DBD::ODBC

On 22/11/2011 08:48, Martin J. Evans wrote:
> On 21/11/11 18:51, Yang, Longjiang wrote:
>> Hi,
>>
>>
>>
>> I use DBD::ODBC to connect to Teradata. I need to get column data type
>> back. when using $sth->{TYPE}. I got IDs not in $dbh->type_info_all. It
>> appears to happen to only CHAR and VARCHAR types. Does anyone know how
>> to get the data type information for a column (in a query)? Thanks.
>>
>>
>>
>> Here is the code. Note that -8 in $sth->{TYPE} is not defined in
>> $dbh->type_info_all. It is the ID for both CHAR and VARCHAR columns
>> defined in the table. I tried the same using DBD::ADO and got similar
>> issue.
>>
>
> type_info_all calls ODBC's SQLGetTypeInfo and the ODBC driver returns 
> the result-set. {TYPE} returns the type returned from SQLDescribeCol. 
> You are obviously on Windows where DBD::ODBC builds in Unicode mode by 
> default. This causes DBD::ODBC to call SQLDescribeColW (the wide 
> version of SQLDescribeCol) and in this case the driver will mostly 
> return -8 (SQL_WCHAR) from char columns. However, DBD::ODBC does not 
> yet call SQLGetTypeInfoW and hence the difference. Arguably, it is a 
> bug in DBD::ODBC so by all means report it at rt.cpan.org and I will 
> fix it as I am working in that area right now.
>
> <snipped test code>
>
> If you are in a position to rebuild a new DBD::ODBC and test the fix 
> that will influence how quickly I get around to making the change.
>
> Martin

In actual fact if you download DBD::ODBC 1.32_3 this version calls 
SQLTablesW in a unicode build so you should try that. Get back to me if 
you find any issues.

Be aware I've not done column_info completely for Unicode yet (and the 
other metadata methods like primary_keys) but I'm working on it right now.

Martin