On Tuesday 19 February 2013 22.18:43 Vishesh Handa wrote: > Hey everyone > > On Wed, Feb 13, 2013 at 8:30 PM, Matthew Dawson <[email protected]>wrote: > > .Especially since you remarked a duplicate is not a big deal, yes you > > should be > > fine. For quick reference of the probabilities, Wikipedia offers: > > > > http://en.wikipedia.org/wiki/Universally_unique_identifier#Random_UUID_pro > > bability_of_duplicates > Considering the probability of hitting a duplicate is so low. I'm going to > be removing the check in master. > > Unless someone objects? > > Very simple benchmarks reveal that the time spent creating a resource drops > from 8 msecs to 4-5sec, when one removes the check. And this is with an > empty database. On a production system, with a large database, the > difference would be a lot greater.
If a proper implementation such as QUUID or boost's uuid is used, this is completely safe (you will never ever hit a duplicate on a single system, and it should even be good enough if you connected all the nepomuk db's of all users together ;-). So, go for it =) Cheers, Christian _______________________________________________ Nepomuk mailing list [email protected] https://mail.kde.org/mailman/listinfo/nepomuk
