[HIVE-611] Support variables for variables for composing data flows

• Type: New Feature
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Component/s: Query Processor
• Assignee: Ashish Thusoo
• Reporter: Ashish Thusoo
• Created: 06/Jul/09 22:19
• Updated: 06/Mar/12 16:03

Description

This is similar to SCOPEs use of variables for composing data flows. With this capability Hive should be able to support things like

X = select c1, c2 from T;
Y = select X.c1 as c1, T2.c1 as c2 from X join T2 on (T2.c2 = X.c2);
Z = select Y.c1, count(1) from Y group by Y.c1

This can be also achieved through views though one minor difference is that the data variables X, Y and Z are local to the session whereas views are global is scope (though you can also have session specific views).

Thoughts?

Sub-Tasks

• variable to be assigned with an atomic value from a query: Open