FreeTDS Development Group

["ZIGLIO, Frediano, VF-IT" <Frediano.Ziglio AT vodafone.com>]

I found some possibles issues on current CVS
- multiple RPCs and cursor updates (ODBC). sql2tds and some function in
odbc_util do not take into account proper row and could lead to some
corruption. Not that used but I think has to be fixed. I discovered this
problem implementing cursors update. To send colunms to server we call
sql2tds using informations from ARD/IRD however sql2tds use some
informations from APD/IPD and using curr_param_row to refer to row
instead of current row however this is wrong for cursor updates (we do
not update curr_param_row here!!!). So I realize that when
sql_desc_octet_length_ptr and/or sql_desc_indicator_ptr are used we
should take into account row, sql_desc_bind_type and
sql_desc_bind_offset_ptr. In odbc_util.c odbc_set_return_status,
odbc_set_return_params and odbc_get_param_len use these members without
considering row so all callers should know which row to refer. Perhaps
we should add and consider row and ARD/APD to these functions... I'll
try to fix in this way...
- autodetect tds version. The problem is that upper library receive all
errors/warning/environment change for all possible connection tries. We
should cache all that things and reissue last try to upper layer
- dblib errors, Microsoft/Sybase. I realize that error constants of the
two library are different!!! This means that libraries are not binary
compatible and perhaps --enable-msdblib can have some problems... The
result is that using MS headers and FreeTDS's libraries can lead to
strange conditions handling errors (errors not handled in the same way).

freddy77