VistaDB 4.3 is an upgrade to VistaDB 4.1 that resolves a number of defects identified in VistaDB 4.x with performance optimizations.
New Features In This Release
Many queries run about 2.5x faster in VistaDB 4.3. The performance improvement comes from several optimizations to the query execution algorithm coupled with result caching within queries. This caching is internal to the VistaDB engine and requires no changes to user code. The caching is sensitive to available memory and will automatically release the least recently accessed cache entries under memory pressure (i.e. when the GC runs).
VistaDB 4.2 introduced a new licensing system that addresses strong feedback from customers
of previous versions of VistaDB.
The new licensing system is a proven part of our Gibraltar product that has worked very well for customers for multiple years.
It requires activation for the developer tools associated with VistaDB, but eliminates the need for .LICX files and supports
activation scenarios even direct internet connectivity is not available.
All existing VistaDB 4.x customers will be emailed new license keys since the new system will not work with license keys generated for previous
versions of VistaDB.
If you are a current VistaDB customer and have not received you new license key, please
Defects Fixed In This Release
This is the list of issues fixed since the last release version (18.104.22.168)
- Fixed: The function IsNumeric() now works correctly for column inputs (evaluates row-by-row) rather than always reporting 0 (false).
- Fixed: The ISNULL() and COALESCE() functions now correctly return non-nullable columns when used in views. This is important in EF applications that require non-nullable key columns.
- Fixed: SP_Rename can now be executed like a procedure call (without parentheses, as in SQL Server) and not just as a function call (both call formats are supported for now).
- Fixed: CREATE INDEX now parses and accepts a [dbo] prefix on the table name (with or without brackets).
- Fixed: CREATE INDEX now parses and accepts the keyword CLUSTERED without throwing an error-however, it still has no effect on the index created.