ODBC DRIVER v06-40-0002 - Mailing list pgsql-interfaces

Hello all,

Here is a MAJOR update to the odbc driver.  Lots of missing
functionality has been added (see below).  There is more to go, but
hopefully this fills in what was lacking, especially in the area of
retrieving data with SQLExtendedFetch and SQLGetData.  I performed
extensive testing (I actually used "odbc test" for the first time).
But, even with all that testing, there might be some bugs, so please let
me know if you find any.

The source code will be in the archive soon, hopefully by tomorrow.
What's currently on our website is just the core odbc code for now.  I
believe I have re-merged Thomas' changes for the Unix version correctly,
so everything should compile ok when it gets there.

Byron



version.txt           date:  12/28/98             version:  06.40.0002
----------------------------------------------------------------------
This file describes changes and or fixes since the last update.

NEW FEATURES:

1. Truncation is now properly implemented when using SQLGetData and
   SQLFetch/ExFetch on character and binary data.  SQLGetData can be
   called multiple times and it will return the correct amount of
   data left.  When all of a column's data has been retrieved, it
   will return SQL_NO_DATA_FOUND, until the row is repositioned, which
   is what the spec. dictates.

2. SQLSetPos (for position/refresh only) is now supported, for both
   Declare/Fetch and regular cursor styles.

3. Full support for column-wise (SQL_BIND_BY_COLUMN) and row-wise
   binding through the statement option SQL_BIND_TYPE.

4. Rowsets greater than 1 are now supported (both Declare/Fetch and
   regular cursor styles) through SQLExtendedFetch.

5. SQLGetData fully supports rowsets greater than 1 in conjunction with
   SQLExtendedFetch, in both Declare/Fetch and regular cursors.

6. All SQLExtendedFetch directions are now supported for regular
   (Non-Declare/Fetch) cursor style, including SQL_FETCH_RELATIVE.
   For declare/fetch, SQL_FETCH_NEXT is now supported, with all
   directions coming soon!

7. All statement options are at least recognized.

8. All connection options are at least recognized.

9. Statement options can be set through a connection using
   SQLSetConnectOption().  These affect all statements on that connection
   and become the default for any future statements created on that
   connection.

10. In-Line Large objects are supported.  Previously, large objects could
    only be created through data-at-execution parameters.  This would have
    prevented applications not using data-at-exec parameters for
    LongVarBinarybinary data from creating any large object (possibly Borland).

11. Type Mapping now spans all ODBC-SQL types.

12. Binary/VarBinary now properly implemented through bytea.  As soon as
    Dave adds the patch for creating bytea with a precision, it will
    hopefully be right on with the spec.

13. Added an experimental driver option (for win32 only) to cause
    SQLCancel to actually call SQLFreeStmt(SQL_CLOSE) in the odbc Driver
    Manager.  This was necessary because some applications call SQLCancel
    and expect it to act like a SQLFreeStmt(CLOSE), which it should, but
    under Windows, the 3.0 driver manager does not recognize it and this
    results in "Invalid Cursor State" errors on some applications.


BUG_FIXES:

1. SQLRowCount returns -1 for unknown, instead of 0.

2. Handles 'Z' in send_function.

3. SQLGetTypeInfo only returns standard SQL types now (no more char2, char4
   stuff).  Each SQL type is now mapped to something.

4. bpchar's name is now "char", which was mandated by #3.

5. Put back in Thomas' changes for mylog() and qlog() with variable
   arguments and other things that got lost.