--- Slepp Lukwai <[EMAIL PROTECTED]> wrote: > On Tue, 9 Oct 2001, Charles Cazabon wrote: > > =>Slepp Lukwai <[EMAIL PROTECTED]> wrote: > =>> > =>> The base question after that comes to be, when tracking a message through > =>> the queue for historical reasons, would using the inode number be a > =>> realistic possibility for a unique identifier > => > =>Inode number is guaranteed to be unique at any given instant. > > So it will eventually be reused, and thereby over a year long period will > more than likely be used many more times than once. Alright. > > =>> or would the message ID be a more appropriate (while harder to find) > =>> item to track? > => > =>Message-IDs are only theoretically unique. In practice, they collide > =>all too often. > > That's unfortunate.. Has anyone ever considered placing in an MTA > Message-ID that can be more or less considered unique among all IDs? I > figure this would be a trivial thing, but I wonder if it has practical > application.
If you consider going to that much trouble, then you might as well just post-process and combine the timestamp of the "new msg" entry. e.g. Oct 9 21:22:26 mail qmail: 1002658946.264686 new msg 147842 could generate a unique ID of 1002658946.264686.147842 or even (coz I dig hashes) 1002658946.264686 new msg 147842 NaN, I know, but they _are_ unique (unless there is a policy of sending the clock back ;). Point is, there is already enough information in the log to make a given entry unique. Adam. ____________________________________________________________ Do You Yahoo!? Get your free @yahoo.co.uk address at http://mail.yahoo.co.uk or your free @yahoo.ie address at http://mail.yahoo.ie
