QVD files can be created and maintained automatically via the buffer prefix. This prefix can be used on most Load and Select (SQL) statements in script. It indicates that QVD Files are used to cache/buffer the result of the statement.
Numerous limitations exist, most notable is that there must be either a file load or a select statement at the core of any complex statement.
The name of the QVD file is a calculated name (a 160-bit hexadecimal hash of the entire following load or select statement and other discriminating info) and is typically stored in the Application Data folder or another folder specified under User Preferences: Locations. This means that the QVD buffer will be rendered invalid by any change in the following load or select statement.
QVD buffers will normally be removed when no longer referenced anywhere throughout a complete script execution in the document that created it or when the document that created it no longer exists. This housekeeping can be turned off by checking the options Keep Unreferenced QVD Buffers in Document Properties: General (first requirement) and Keep Orphaned QVD Buffers in User Preferences (second requirement) property pages respectively, though the procedure cannot be generally recommended.
The syntax is:
buffer[ (option [ , option] ) ] ( loadstatement | selectstatement )
option ::= incremental | expiry
expiry::= stale [after]amount[ (days | hours)]
amount is a number specifying the time period. Decimals may be used. The unit is assumed to be days if omitted.
The incremental option enables the ability to read only part of an underlying file. Previous size of the file is stored in the XML header in the QVD file. This is particularly useful with log files. All records loaded at a previous occasion are read from the QVD file whereas the following new records are read from the original source and finally an updated QVD-file is created. Note that the incremental option can only be used with load statements and text files and that incremental load cannot be used where old data is changed or deleted!
The stale after option is typically used with DB sources where there is no simple timestamp on the original data. Instead one specifies how old the QVD snapshot can be to be used. A stale after clause simply states a time period from the creation time of the QVD buffer after which it will no longer be considered valid. Before that time the QVD buffer will be used as source for data and after that the original data source will be used. The QVD buffer file will then automatically be updated and a new period starts.
If no option is used, the QVD buffer created by the first execution of the script will be used indefinitely.