Hi Frank,
I'd like to share a few thoughts that occured to me while I was under the shower (it always happens so for me, I don't know for you...). The KDE address book driver is split into two parts: - one DLL that loads the second DLL if certain conditions are met (namely, KDE at run time is present and not too old nor too recent) - a second DLL that is linked against relevant KDE libraries and that does the real work. The same occurs to other DBA drivers; they are split in two, and only the second part is linked against the needed libraries. Some drivers should be split that way, but it's not done yet. So this framework is more or less general. I think that, if we redesign the drivers some day, it would be more efficient to group all "first part" (the leightweight ones) into one single library that decides the loading of the "real" drivers at run time. That would reduce the number of DLLs installed, and probably result in less disk I/O too. The only drawback I can imagine is that the common part will mix code from different origins, in other words there would be less modularity of code. If I am wrong, please tell me why. Just my two cents... -- - Maman, quel est le cri de la fourmi ? - La fourmi croonde, mon chéri. --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
