On May 7, 2012, at 3:44 PM, Richard Hipp wrote: > Call it whichever you want. SQLite has *always* worked this way, for > reasons that go very deeply into the design. And applications exists which > depends on the current behavior. So it probably isn't going to change.
Primum non nocere. Rather unsettling in that specific scenario as an update ends up deleting half of the existing rows :( In any case, what about supporting DEFERRABLE INITIALLY DEFERRED for unique constraints? That would nicely sort this out, and provide a very convenient feature, on par with the one for referential constraints. Thoughts? _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users