On 6/13/17, Scott Robison <sc...@casaderobison.com> wrote: > > Is it fair to say that the rowid aliasing behavior does not require > (by design) the incantation "INTEGER PRIMARY KEY" (all three words in > that order as the "type") as long as the type is INTEGER and the > constraint PRIMARY KEY appears somewhere in the column's constraint > list?
See https://www.sqlite.org/lang_createtable.html#rowid CREATE TABLE t(x INTEGER PRIMARY KEY DESC, y, z); Columns x *not* a ROWID. Life is complicated. Sorry for the mess. But we have to preserve backwards compatibility - even with bugs like this. > -- > Scott Robison > _______________________________________________ > sqlite-users mailing list > sqlite-users@mailinglists.sqlite.org > http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users > -- D. Richard Hipp d...@sqlite.org _______________________________________________ sqlite-users mailing list sqlite-users@mailinglists.sqlite.org http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users