Column Reference

A column object belongs to a table and holds a sequence of data. To refer to a column object, we can use <table>.<column> if the statement is not a SQL statement. Please note that <table>.<column> is read only and cannot be used to modify a column.

t=table(2 3 4 as x, 5 6 7 as y);
t;


x	y
2	5
3	6
4	7

t.x;
// output: [2,3,4]

a=t.x;
a+=1;
a;
// output: [3,4,5]

t.x;
// output: [2,3,4]
// column t.x is unchanged

t.x+=1;
// output: Syntax Error: [line #1] Can't use assignment in expression.
// this is because t.x is read only. To modify column x, use t[`x] or update clause.

In a SQL statement, the table name can be ignored in front of the variable name.

// output: select x from t;


x
2
3
4

How does the system tell if a variable is column reference, a variable name or a function name? This problem is dynamically disambiguated at runtime with the order of column reference, variable name, function name. Specifically, in a SQL statement, if a table column and a defined variable have the same name, the system always interprets the name as the table column. If we want to refer the defined variable in a SQL statement, we have to rename the variable. If a function has the same name as a table column or a variable and we want to use the function name in a SQL statement, we can use an "&" or module name before the function name to qualify it.

x=1..3
z=7..9
t1=select x, y, z from t;

t1;
// since table t has column x, the system interprets x as t.x in the sql statement although there is a separately defined variable x.
// although z is not in t1, the system recognizes it is of the same size as t.x and t.y. Therefore the system uses z together with t.x and t.y to make a new table t1.


x	y	z
2	5	7
3	6	8
4	7	9

In the following example, there is a column, a variable, and a function all named avg. The system resolves the name in the following order: it first attempts to resolve it as t.avg; if that fails, it tries to resolve it as the variable avg; if that also fails, it resolves it as the function. As a result, the first parameter of contextby is resolved as the function, while the others are resolved as t.avg. Users can also explicitly specify &avg or ::avg to ensure it is resolved as the function.

avg=12.5
t1=table(1 2 2 3 3 as id, 1.5 1.8 3.2 1.7 2.5 as avg)
select * from t1 where avg > contextby(avg,avg,id);
// select * from t1 where avg > contextby(&avg,avg,id);
// select * from t1 where avg > contextby(::avg,avg,id);


id	value
2	3.2
3	2.5


t=table(1 2 3 as ID, 4 5 6 as check)

check=1
check1=10
def check(x):x*10
def check1(x):x*100
def check2(x,y):x+2*y;

t;


ID	check
1	4
2	5
3	6

select ID+check from t;
// column ID plus column check from table t


add_ID
5
7
9

select ID+check1 from t;
// column ID plus the variable of check1


add_ID
11
12
13

select accumulate(check2, check) as y from t;
// apply the accumulate template on the function definition of check2 and the column check from table t


y
4
14
26