If a script has no DUMP or STORE statement, there's nothing for pig to execute. This is totally obvious once you understand pig, but can be baffling to a novice.
Pig should emit a warning: "Nothing to execute. DUMP and STORE cause Pig to execute a plan; ILLUSTRATE, EXPLAIN and DESCRIBE illuminate what it would execute. Did you mean to add one of those?" if nothing would cause output.
There's a good argument for having it error with a non-zero exit code (rather than warn), but it is reasonable to run without content for linting purposes, so I believe an ignorable warning is best.
The presence of an ILLUSTRATE, EXPLAIN or DESCRIBE should prevent the warning; it's pretty typical to comment out the store statement in favor of one of those while debugging.
A pig script with only shell statements is a tougher call... Pig would be about the worst possible choice for a sysadmin scripting language, but I suppose there are oozie flows out there that might have Pig scripts with a shell statement. I believe that if there is a logical plan but no statement to cause output Pig should warn; a script of only shell statements should not cause a warning.
I'm happy to implement this – my guess is that I'd have to
1. find the point where pig in script mode decides what to execute;
2. query the ?logical? plan about whether it includes one or more DUMP, STORE, ILLUSTRATE, EXPLAIN or DESCRIBE statements; and
3. issue a log line at WARN level if none were found
If someone can provide any brief high-level guidance about where to put the decision and how to query the plan, I'd be grateful.