In some email I received from Magnus Sundberg <[EMAIL PROTECTED]> on Wed, 02 Jul 2003 10:49:13 +0200, wrote:
> <snip> > > I beleive Gianni has a point, even though I run mysql. > Let's take the IMAP server as an example. > 1. Parse IMAP message > 2. Call C-function that defines the IMAP command. > 3. The C function calls the stored procedure , when applicable > (PostgreSQL) or performs the required SQL commands (MySQL). > 4. Return result Magnus, all this sounds great, that means the codebase will create a barrier between both APIs also a gap which will limit the whole idea for one generic API, which would save time and resources writing more code and more code and more code.. I think the bare optimization limit should be in things like indeces/constraints/queries/code other things like 'db optimization' should be performed outside this circle, i.e. by the user himself, like tweaking PgSQL conf file rather than writing pl/c functions. > But what is the performance gain? Is it worth the effort. I do > beleive this is the more elegant way to implement the database > access, but we will get two quite different code bases. Yes, obviously there might be a performance gain, at least using transactions, I dont agree using stored procedures/functions for a simple reason - portability. > I also beleive it is better to implement these changes, if they > are necessay as soon as possible. Hope this makes some sense. cheers
