Sorry, hit send too fast...As a work around for now, what I ended up doing was using the 'sed' command to replace all of my variables with a shell script. Kind of a hack, but it worked in the end for what I needed.
Carla -----Original Message----- From: Staeben Carla (Nokia-LC/Boston) Sent: Friday, March 30, 2012 11:30 To: user@hive.apache.org Subject: RE: Variable Substitution Depth Limit Yeah, you're not the only one who's run into that issue. There is an open Jira item for it, so they're aware we'd like it configurable anyway... https://issues.apache.org/jira/browse/HIVE-2021 -----Original Message----- From: ext Edward Capriolo [mailto:edlinuxg...@gmail.com] Sent: Friday, March 30, 2012 11:19 To: user@hive.apache.org Subject: Re: Variable Substitution Depth Limit Unfortunately you are going to have to roll your own hive. It was just a concept we borrowed from Hadoop since it does not support more then 40 depth substitution. We can probably make it configurable by a hive-site property. On Fri, Mar 30, 2012 at 10:59 AM, Tucker, Matt <matt.tuc...@disney.com> wrote: > I'm trying to modify a script to allow for more code reuse, by > prepending table names with a variable. > > > > For example: CREATE TABLE etl_${hiveconf:table}_traffic AS . > > > > The problem I'm running into is that after building all of these etl_* > tables, I use a final query to join all of the tables and select all > of the appropriate columns. In this query, I reference the > ${hiveconf:table} variable 122 times, and end up receiving this error: > "java.lang.IllegalStateException: Variable substitution depth too large: > 40". According to a quick Google search, it appears that the limit of > 40 is hardcoded into the VariableSubstitution class. > > > > Is there any way to override this in a hiveconf variable? > > > > Thanks > > > > Matt Tucker > >
