And there are situations in which such a network-oriented scheme is an excellent idea. I do have a quibble. I suspect that an implementation
"by figuring out the most recent modification date/time among all the chunks of the file, across all the relevant servers, and comparing that with the time-to-live value." should be replaced by the testing of a last|latest update-time value that is replaced unconditionally every time an update operation ends successfully. John Gilmore, Ashland, MA 01721 - USA ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
