On Thu, Sep 25, 2014 at 4:46 PM, Mark Lawrence <no...@null.net> wrote:
> > If you are going to keep this behaviour would it not make more sense to > ensure that the table creation fails? The DEFAULT clause is pretty > straight-forward and I don't find it intuitive to go looking for > PRIMARY KEY documentation when it is ignored. > > SQLite should either fail to accept the statement or do what the table > definition says - anything else means heartache for those debugging > problems (which I did for hours on this issue) and those reading the > code afterwards. > That would break backwards compatibility for the millions and millions of applications currently using SQLite. Most of those millions would be unaffected, no doubt, but out of millions I'm sure there are a goodly number that would break. I am unwilling to implement a breaking change simply to make the interface more "intuitive". Had you brought this up in 2002, the outcome would likely have been very different. But at this point, the behavior of INTEGER PRIMARY KEY in SQLite is not something that can be modified. Sorry to disappoint. -- D. Richard Hipp d...@sqlite.org _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
