236019 – [compiler] Compiler doesn't complain for a function which returns an int, returns indeed null

Bug 236019 - [compiler] Compiler doesn't complain for a function which returns an int, returns indeed null

Summary: [compiler] Compiler doesn't complain for a function which returns an int, ret...

Status:	RESOLVED INVALID

Alias:	None

Product:	JDT
Classification:	Eclipse Project
Component:	Core (show other bugs)
Version:	3.3.2
Hardware:	PC Windows XP

Importance:	P3 normal (vote)
Target Milestone:	3.4
Assignee:	Philipe Mulet
QA Contact:

URL:
Whiteboard:
Keywords:

Depends on:
Blocks:

Reported:	2008-06-06 07:32 EDT by David Pérez
Modified:	2008-06-11 04:36 EDT (History)
CC List:	2 users (show)

See Also:

Attachments
Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.

Description David Pérez

2008-06-06 07:32:15 EDT

Build ID:  M20080221-1800

Steps To Reproduce:
1.Create a Java project
2.Create a .java source file
3.Enter a function like this:

class Test {
    ArrayList params;

    public int getSqlParamCount() {
        return params == null ? null:params.size();
    }
}


More information:
If we remove the conditional, then the compiler complains:

class Test {
    ArrayList params;

    public int getSqlParamCount() {
        return null;
    }
}

Comment 1 Olivier Thomann

2008-06-06 09:35:44 EDT

Are you compiling in 1.5 mode ?
What are your compiler settings?
In 1.5, I think autoboxing is used in this case.
It leads to a NPE at runtime.

Comment 2 Olivier Thomann

2008-06-06 09:36:15 EDT

javac 1.6 and 1.5.0 also compile this code.

Comment 3 David Pérez

2008-06-06 10:28:34 EDT

That's true, I'm using 1.6 target.

Even though, javac behaves so, I would consider this behavior as a bug.

Comment 4 Philipe Mulet

2008-06-10 18:34:13 EDT

The conditional operator offers special type checking rules, which allow for a boxing conversion to occur for free. The plain 'return null' is not as sophisticated.

This is a situation where we stick to the language spec, and you should blame the spec (and then we would happily tune our semantics accordingly if you win this fight).

Comment 5 Philipe Mulet

2008-06-10 18:35:14 EDT

Added AutoboxingTest#test159

Comment 6 David Pérez

2008-06-11 04:36:34 EDT

Thanks for the explanation.
Luckily this behaviour isn't so important, so I'm not going to fight against anyone for this.