 |
||
 |
||
|
||||
Table of Contents
Secondly, we could have either passed an extra flag in between the files from
the parser to the handler and finally to the ha_berkeley
class. This would have caused us to modify these classes to accommodate one
extra bit of information and probably some unnecessary information would have
been stored in dependant structures/classes. The other alternative, which we
implemented, was to use a common header file between all three of these to store
the fact whether the table being created is temporal or non-temporal. The lifetime
of this variable goes from the parser to the creation of the TABLE
structure and ends after the underlying creation of the table in BerkeleyDB.
The advantages of this method are minimum modification of code and ease of implementation.
Whenever a Transaction-Time table is being created using the specified syntax, the underlying BerkeleyDB
in instructed to create a table (BerkeleyDB calls this a 'database') with the added functionality. These functions
can be used by the ha_berkeley class to execute various SQL commands.
Developed by the TAU Project, Computer Science Department, University of Arizona