Martin Simmons wrote: >>>>>> On Thu, 16 Apr 2009 10:33:43 +0800, Craig Ringer said: >> In the above, mtime is defined as a database `timestamp' rather than >> stored as the raw integer value from the source host. Note that the >> database may not store timestamps with the same precision as the fd >> host, and may store them using an internal floating-point form, so some >> error must be allowed for during verification. > > Inaccuracy in mtime and ctime would break the accurate backups code.
It's sub-millisecond error. It just requires that comparisions for equality be done in a similar way to floating-point comparisons, eg: a < b + err && a > b - err which may or may not be a performance issue, but isn't a functional one. It's one of the main things I want to look at, since for many other purposes having usable timestamps in the database would be REALLY useful. If it proves necessary to store binary-exact mtime and ctime values in the DB, some job-level information on client timestamp format and timezone might at least allow functions in the database to usefully interpret these values (and, with functional indexes, search them quickly albeit with an insert performance cost). -- Craig Ringer ------------------------------------------------------------------------------ Stay on top of everything new and different, both inside and around Java (TM) technology - register by April 22, and save $200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco. 300 plus technical and hands-on sessions. Register today. Use priority code J9JMT32. http://p.sf.net/sfu/p _______________________________________________ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
