I think keeping the null value will be better, because it's consistent with relational database.
Best, -------------- Yuan Tian On Mon, Jul 18, 2022 at 2:43 PM 廖兰宇 <[email protected]> wrote: > > Hi everybody, > > In 0.13/0.12, when you execute query sql "select a from root.sg where b >= > 3 " with the following dataset: > +-------------------------------------+---------+---------+ > | > > Time|root.sg.a|root.sg.b| > +-------------------------------------+---------+---------+ > |1970-01-01T08:00:00.001+08:00| > 1| null| > |1970-01-01T08:00:00.002+08:00| > 2| null| > |1970-01-01T08:00:00.003+08:00| null| > 3| > |1970-01-01T08:00:00.004+08:00| null| > 4| > +-------------------------------------+----------+--------+ > The expected result set would be empty because the values of root.sg.a are > null at timestamp 3 and 4(b >= 3). > +----+----------------+ > |Time|root.test.test.a| > +----+----------------+ > +----+----------------+ > Empty set. > We suggest keeping the null value of the columns at the timestamp given by > filter. You can use "select a from root.sg where b >= 3 and a is not null" > if you do not want to keep the null value. > +-------------------------------------+---------+ > | > > Time|root.sg.a| > +-------------------------------------+---------+ > |1970-01-01T08:00:00.003+08:00| null| > |1970-01-01T08:00:00.004+08:00| null| > +-------------------------------------+---------+ > Any comments would be much appreciated. > > Best, > > ------------------------------------ > > Lanyu Liao > > School of Software, Tsinghua University > > 廖兰宇 > > 清华大学 软件学院
