nick huang wrote: > You are absolutely right that its type is NULL and min(id) returns 0 as I use > sqlite_column_int64 to retrieve data. However, > isn't it confusing that the "sqlite3_step" returns SQLITE_ROW which usually > means some dataset is returned?
A resultset consisting of one row is indeed being returned. What's confusing about this? > I mean, if it returns > one NULL row, why doesn't sqlite simply return "SQLITE_DONE" or something > instead of "SQLITE_ROW"? An empty resultset and a resultset with one row that just happens to have NULLs in all columns are two entirely different things. Compare and contrast: select count(*) from (select id from url); select count(*) from (select min(id) from url); > You see, as user I have to > double-check its return type even after I get "SQLITE_ROW". In this sense, I > think it is a bug. Can you perhaps point to a clause in SQL-92 standard that SQLite violates here? A query involving an aggregate function without a GROUP BY clause always returns exactly one row. This is how SQL works. Igor Tandetnik _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
