Re: ODBC Driver

Il sab, 2003-03-22 alle 02:41, Peter Deacon ha scritto: 
> Hi there.

Hi Peter.

> Grabed the latest freetds snapshot yesterday.  In a few hours I managed to
> get ODBC driver working darn near perfect... Thought for sure it would
> take weeks of hacking before I'd even be able to login let alone do
> anything useful.
> 
> We've played with several third party ODBC drivers for MSSQL
> and Sybase over the years and I've gotta say from what I've seen so far
> FreeTDS is second only to Merant/Data direct/ whatever their called today.

Thanks

> Best of all, there was already a visual studio project in the tarball..
> all I had to do was compile.. Nothing like being able to step from
> your application right into the driver (Edit and continue anyone?:)
> 
> Heres what I've found so far..
> 
> When we call stored procedures have a habbit of using caps to spell out
> CALL.  The prepare routine would fail if not all lowercase 'call'.

Fixed in CVS

> For SQLGetDiagField
> Implemented:
> 	SQL_DIAG_ROW_COUNT
> 	SQL_DIAG_CLASS_ORIGIN
> 	SQL_DIAG_SS_MSGSTATE (The SS functions I set win32 only)
> 	SQL_DIAG_SS_LINE
> 	SQL_DIAG_NATIVE
> 	SQL_DIAG_SQLSTATE

Good!

> numRecord should be decremented after NO_DATA_FOUND check.

Fixed in CVS (with another problem...).

> Misc:
> Return 'Microsoft SQL Server' if using MSSQL from SQLGetInfo

Fixed in CVS.

> With text datatypes there was an off by one.. It would move the value
> pointer back one making the first letter garbage and exclude the
> last.

I didn't understand...

> I've always wondered about where SQLStates come from.  Sybase stores a few
> in their sysmessages table however as far as I can tell they don't exist
> in the MS sysmessages table.   Cought the bits in freetds where it can
> fetch sqlstates from the wire..but it never appears to get called.  If
> sqlstate is not otherwise set added lookup tables for MSSQL and Sybase
> which map about 150 native errors to SQL States..  Hope this sounds
> reasonable?

No, simply sql server return sql state itself. Perhaps libtds discard
this information... Another problem is translation from odbc2 codes to
odbc3 codes...

> Anyway have a few weird bits left to check into before posting my diff.

Well, post your diff.

> You have no idea how cool this is :)
> 
> Thanks everyone
> -Peter

  Frediano Ziglio