On Mon, Jun 18, 2007, Paul J Stevens <[EMAIL PROTECTED]> said: > Aaron Stone wrote: >> Oh my, I never realized that the header names were repeated for each >> physmessage. Of course that should be changed -- sounds like Paul is >> already working on it. > > Not yet, actually :-). Been keeping busy with the suppress_duplicate, > idle, and mimechunking branches. > > And it's not the headernames that are stored repeatedly, but the > headervalues. Headernames are already unique. > > Storing unique headervalues is in the same order of impact as the > mimechunking change. It'll take some time to mull over before I'll be > able to do it right.
Oh, ok, cool, then it is what I thought it was in the first place :-) I'm not convinced that there's much point to having another bridge table for the headers. We're not talking about very much data, and I'd be concerned that we would be killing off some useful indexes for fast header retrievals. Aaron _______________________________________________ DBmail mailing list [email protected] https://mailman.fastxs.nl/mailman/listinfo/dbmail
