On Fri, Sep 6, 2013 at 2:50 PM, Marc L. Allen <mlal...@outsitenetworks.com>wrote:
> No one commented on my second thread (written after I actually understood > the problem!). > > But, I proposed a two update sequence to do it. > > UPDATE table SET Sequence = -(Sequence + 1) WHERE Sequence >= > seq_to_insert AND Name = name_to_insert > UPDATE table SET Sequence = -Sequence WHERE Sequence < 0 AND Name = > name_to_insert > Because, no offense, that's just a work-around. The question that matters is whether SQL requires the UPDATE statement to succeed or not, despite a given implementation possibly having to deal with transient index violations while processing the rows. If so, SQLite would be non-compliant in that regard, as of now. If not, and implementations are free to support or not such UPDATE statements, then SQLite would ideally document this limitation, since it departs from major DBMS's like Oracle. Myself, if I'm "thinking in sets", all implementation details aside, the UPDATE statement looks fine and correct, and I'd have expected SQLite to support it. But I'm just waiting to read Dr. Hipp's own read on this now. --DD _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
