psqlODBC Configuration Options

Advanced Options (Driver) Dialog Box

• DEFAULTS: Press to this button restore the normal defaults for the settings described below.
   
• Disable Genetic Optimizer: Automatically turns off the backend genetic optimizer at connection time. This is a convenience feature rather than having to type it in the connect settings. This feature was added when we noticed the backend seemed to have big problems optimizing some queries.
   
• KSQO (Keyset Query Optimization): This feature helps certain queries run in the backend without crashing. Some applications, specifically the MS Jet Database Engine, use "keyset" queries of the form:
  
  select...where (a = 1 AND b = 1 AND c = 1) OR (a=1 AND b=1 AND c = 2)...
  
  These types of queries will most likely crash the backend without the KSQO feature.
   
• CommLog (C:\psqlodbc.log): Log communications to/from the backend to that file. This is good for debugging problems.
   
• Recognize Unique Indexes: This controls what SQLStatistics() will return about unique indexes. By default it is set to false. This setting allows Access 95 and 97 to ask the user at link time what the index should be. This is meant as a workaround until the backend canonifier can handle the Microsoft Access queries better. For some applications (i.e., Access 2.0), you must set this option to update any tables.
   
• ReadOnly (default): New datasources will inherit the state of this box for the datasource readonly attribute.
   
• Use Declare/Fetch: If true, the driver automatically uses declare cursor/fetch to handle SELECT statements and keeps 100 rows in a cache. This is mostly a great advantage, especially if you are only interested in reading and not updating. It results in the driver not sucking down lots of memory to buffer the entire result set. If set to false, cursors will not be used and the driver will retrieve the entire result set. For very large tables, this is very inefficient and may use up all the Windows memory/resources. However, it may handle updates better since the tables are not kept open, as they are when using cursors. This was the style of the old podbc32 driver. However, the behavior of the memory allocation is much improved so even when not using cursors, performance should at least be better than the old podbc32.
   
• Parse Statements: If enabled, the driver will parse an SQL query statement to identify the columns and tables and gather statistics about them such as precision, nullability, aliases, etc. It then reports this information in SQLDescribeCol, SQLColAttributes, and SQLNumResultCols. Prior to PostgreSQL 6.4, this was the only accurate way of getting information on precision from a query result.
  
  If the parser can not deal with a column (because it is a function or expression, etc.), it will fallback to executing the statement which is the old style of getting the info. The parser is fairly sophisticated and can handle many things such as column and table aliases, quoted identifiers, literals, joins, cross-products, etc. It can correctly identify a function or expression column, regardless of the complexity, but it does not attempt to determine the data type or precision of these columns.
   
• Unknown Sizes: This controls what SQLDescribeCol and SQLColAttributes will return as to precision for character data types (varchar, text, and unknown) in a result set when the precision is unknown. This was more of a workaround for pre-6.4 versions of PostgreSQL not being able to return the defined column width of the varchar data type.
  
  
  • Maximum: Always return the maximum precision of the data type.
  • Dont Know: Return "Don't Know" value and let application decide.
  • Longest: Return the longest string length of the column of any row. Beware of this setting when using cursors because the cache size may not be a good representation of the longest column in the cache.
  
  MS Access: Seems to handle Maximum setting ok, as well as all the others.
  Borland: If sizes are large and lots of columns, Borland may crash badly (it doesn't seem to handle memory allocation well) if using Maximum size.
  
  
• Data Type Options: affects how some data types are mapped:
   
• Text as LongVarChar: PostgreSQL TEXT type is mapped to SQLLongVarchar, otherwise SQLVarchar.
• Unknowns as LongVarChar: Unknown types (arrays, etc) are mapped to SQLLongVarChar, otherwise SQLVarchar
• Bools as Char: Bools are mapped to SQL_CHAR, otherwise to SQL_BIT.


• Cache Size:When using cursors, this is the row size of the tuple cache. If not using cursors, this is how many tuples to allocate memory for at any given time. The default is 100 rows for either case.
   
• Max Varchar The maximum precision of the Varchar and BPChar(char[x]) types. The default is 254 which actually means 255 because of the null terminator. Note, if you set this value higher than 254, Access will not let you index on varchar columns!
   
• Max LongVarChar: The maximum precision of the LongVarChar type. The default is 4094 which actually means 4095 with the null terminator. You can even specify (-4) for this size, which is the odbc SQL_NO_TOTAL value.
   
• SysTable Prefixes: The additional prefixes of table names to regard as System Tables. The driver already treats names that begin with "pg_" as system tables. Here you can add additional ones, such as data dictionary tables (dd_). Separate each prefix with a semicolon (;)
   
• Connect Settings: These commands will be sent to the backend upon a successful connection. Use a semi-colon (;) to separate commands. This can now handle any query, even if it returns results. The results will be thrown away however!
   

Advanced Options (DataSource or Connection) Dialog Box

• ReadOnly: Whether the datasource will allow updates. For new datsources, the default value comes from the Advanced Options (Driver) dialog box.
   
• Connect Settings: The driver sends these commands to the backend upon a successful connection.  It sends these settings AFTER it sends the driver "Connect Settings". Use a semi-colon (;) to separate commands. This can now handle any query, even if it returns results. The results will be thrown away however!
   
• Row Versioning: Allows applications to detect whether data has been modified by other users while you are attempting to update a row. It also speeds the update process since every single column does not need to be specified in the where clause to update a row. The driver uses the "xmin" system field of PostgreSQL to allow for row versioning. Microsoft products seem to use this option well. See the faq for details on what you need to do to your database to allow for the row versioning feature to be used.
   
• Disallow Premature: Disallow Premature is an option to compensate for the lack of a server's functionality. For example (Middleware) applications issue the following ODBC API calls.
  
  SQLPreapare(hstmt, "select ...", ..)
  
  In most cases they have to know how many fields, what kind of fields they would return and so they would issue
  
  SQLNumResultCols and/or
  SQLDescribeCols/SQLColAttribute etc.
  
  The problem is how the psqlODBC driver answers the inquiry. PostgreSQL hasn't provided the Prepare functionality yet and we couldn't ask the backend about it directly. When using Disallow Premature, the driver will get the column info as follows:
  
     begin;(unless in a transaction)
     declare cursor .. for select ...
     fetch backward in ..
     close ..
  
  The driver gets the field info using the fetch backward's result. The fetch backward command returns no row but returns the field info. I also expected the the fetch backward command returns immediately but unfortunately it isn't always true. The 7.1 or later servers do seem to return from the fetch backward command immediately.
   
• Show System Tables: The driver will treat system tables as regular tables in SQLTables. This is good for Access so you can see system tables.
   
• OID Options:
   
• Show Column - Includes the OID in SQLColumns. This is good for using as a unique identifier to update records if no good key exists OR if the key has many parts, which blows up the backend.
   
• Fake Index - This option fakes a unique index on OID. This is useful when there is not a real unique index on OID and for apps which can't ask what the unique identifier should be (i.e, Access 2.0).




• Protocol (note that the protocol option has no effect from version 08.01.0001 onwards):
   
• 6.2: Forces driver to use PostgreSQL 6.2 protocol, which had different byte ordering, protocol, and other semantics.
   
• 6.3: Use the 6.3 protocol. This is compatible with both 6.3 and 6.4 backends.
   
• 6.4: Use the 6.4 protocol. This is only compatible with 6.4 backends.