FreeTDS Development Group

[Frediano Ziglio <freddyz77 AT tin.it>]

Il lun, 2004-05-10 alle 19:54, Piet De Jong ha scritto:
> Hi All,
> Attached I have a very simple select statement which is supposed to
> return a single column.
> However when I run the program I do not get any data returned.
>  
> I use the latest version of OTL ( 4.0.84 )
>  
> I am not sure if I had asked this question before, I could not find it
> on the list.
>  
> The platform with the problem: 
> Slackware 9.1.0
> using iodbc 3.51.1
> gcc version: 3.2.3
>  
> When I compile and run the attached example on Windows I dont have any
> problems, the correct values ( 1  and RPC = 1 ) are returned.
>  
> Many thanks
> PIet
>  

>From log (unixODBC)

[ODBC][21603][SQLPrepare.c][173]
                Entry:
                        Statement = 0x91ce160
                        SQL = [select revision from revisiontable where
version = 1][length = 52 (SQL_NTS)]
[ODBC][21603][SQLPrepare.c][374]
                Exit:[SQL_SUCCESS]
[ODBC][21603][SQLNumResultCols.c][142]
                Entry:
                        Statement = 0x91ce160
                        Column Count = 0xbfeea3ac
[ODBC][21603][SQLNumResultCols.c][233]
                Exit:[SQL_SUCCESS]
                        Count = 0xbfeea3ac -> 0
[ODBC][21603][SQLSetStmtAttr.c][217]
                Entry:
                        Statement = 0x91ce160
                        Attribute = SQL_ATTR_ROW_ARRAY_SIZE
                        Value = 0x1
                        StrLen = -3
[ODBC][21603][SQLSetStmtAttr.c][666]
                Exit:[SQL_SUCCESS]
[ODBC][21603][SQLSetStmtAttr.c][217]
                Entry:
                        Statement = 0x91ce160
                        Attribute = SQL_ATTR_ROWS_FETCHED_PTR
                        Value = 0x91ce01c
                        StrLen = -3
[ODBC][21603][SQLSetStmtAttr.c][666]
                Exit:[SQL_SUCCESS]
[ODBC][21603][SQLExecute.c][186]
                Entry:
                        Statement = 0x91ce160
[ODBC][21603][SQLExecute.c][341]
                Exit:[SQL_SUCCESS]
[ODBC][21603][SQLFetchScroll.c][145]
                Entry:
                        Statement = 0x91ce160
                        Fetch Orentation = 1
                        Fetch Offset = 1
[ODBC][21603][SQLFetchScroll.c][346]
                Exit:[SQL_SUCCESS]
[ODBC][21603][SQLFetchScroll.c][145]
                Entry:
                        Statement = 0x91ce160
                        Fetch Orentation = 1
                        Fetch Offset = 1
[ODBC][21603][SQLFetchScroll.c][346]
                Exit:[SQL_NO_DATA]
[ODBC][21603][SQLFreeStmt.c][133]
                Entry:
                        Statement = 0x91ce160
                        Option = 0
[ODBC][21603][SQLFreeStmt.c][246]
                Exit:[SQL_SUCCESS]
[ODBC][21603][SQLFreeStmt.c][133]
                Entry:
                        Statement = 0x91ce160
                        Option = 1
[ODBC][21603][SQLFreeHandle.c][353]
                Entry:
                        Handle Type = 3
                        Input Handle = 0x91ce160
[ODBC][21603][SQLFreeHandle.c][454]
                Exit:[SQL_SUCCESS]
[ODBC][21603][SQLEndTran.c][315]
                Entry:
                        Connection = 0x91a7dd0
                        Completion Type = 0

The key it's always (like Oracle connector) SQLNumResultCols after
SQLPrepare... trying with --enable-developing you will get what you
expect:

$ rm /tmp/sql.log && g++ test_select.cpp -lodbc && ./a.out
test_select.cpp:78 - Rev [1]
test_select.cpp:79 - RPC: 1

Current implementation it's no so instable however performance are not
that fine...

More deeply from code comments:

 /* try to prepare query */
 /* TODO try to prepare only getting informations (faster and
optimizable) for mssql2k */

This can wait however mssql2k can prepare and execute using a single
round-trip (a single store procedure call). Not stopping (for 0.63)

 /* TODO support getting info for RPC */

I don't even know if this is possible...

 /*
  * TODO this should be done if all parameters are bounded or using TDS5
  * binary_test for example do not work for this problem (post bind on
image)
  */

binary_test now works using Sybase, not a stop-over, however this
optimization IMO should be implemented in 0.63!

 /* TODO use current parameter informations */
 stmt->need_reprepare = 1;

quite bad... for every prepare it prepare statement 2 times... just
cause it need to pass types...


This just to explain that this way of getting result it's not that
good!!! IMHO library should change their code to a more optimized way!!!

freddy77