|
From: Kenneth D. <ke...@se...> - 2009-05-02 14:21:49
|
While I'm in question-everything mode, one thing I've always wanted for Andromeda is virtual columns. The idea is this. I've got a table with some very complex calculations. Every intermediate step of every calc is embedded in the table as a value. On the plus side, this lets people use the Andromeda philosophy of "debugging the data", which is cool. On the down-side, the performance of write operations is directly linear to the row size, so Andromeda apps are always slower to insert or update. The more you get the Andromeda philosophy, the slower your app is. Very bad. Now, in transactional biz apps, this doesn't tend to matter in the end. The difference between a 50ms insert and a 100ms insert is nothing to the user. But still, wouldn't it be better to let the programmer decide which columns he wants to see, and which to hide for the sake of performance? So we get the idea of a "virtual" column. These are meant to be used in calculations where you do not want to see the intermediate steps. The column's value is generated and used in triggers to perform calculations, but it is not part of the table, the value disappears when the trigger is finished. It is a temp variable. Just thinking out loud... -- Kenneth Downs Secure Data Software ke...@se... www.andromeda-project.org www.secdat.com Office: 631-689-7200 Cell: 631-379-0010 Fax: 631-689-0527 |