From: Tapani T. <tt...@it...> - 2003-05-02 10:41:42
|
On Fri, May 02, 2003 at 09:56:33AM +0100, Lars Hecking wrote: > My plan is to let the clients pass the temp file name to the daemon, > not just the directory name, and also let the clients rather than the > daemon delete the temp directory (let each component remove only the > file/dirs it creates). That sounds like a good idea, even as a general principle. > My other plan is to add smtp and exec capabilities to the non-milter > client, so that all mail forwarding code is removed from the daemon. > In principle, this could make daemon-client communication very simple > (infected, not infected), although the daemon still needs envelope > info for notifications. Hmm. I'd also like to have the daemon make decisions about adding header(s), adding or removing recipients (like, maybe we'd want to forward infected messages to a special host elsewhere), or even replacing the body (after removing the virus, should someone want to try that). With a suitable protocol it should be possible for the daemon to be more or less ignorant about whether it's talking to milter or a stand-alone client. -- Tapani Tarvainen |