huge news system architecture

[Katsuhiro Kondou]

In article <20020225015340.GB16549 at wonderland.linux.it>,
	"Marco d'Itri" <md at Linux.IT> wrote;

}  >should be already gone.  I've considered a storage method
}  >which utilizes remote server thru nntp, but I couldn't
}  >reach good way to expire overview data in reasonable time.
} What's wrong with processing cancels?

It should be fine if canceling message only by SMcancel.
How does it know deleted messages manually or by system
crash? 

And it's necessary to consider history/overview recreation
from the remote spool thru nntp.  I dont't think it could
be done in reasonable time.
-- 
Katsuhiro Kondou