SpEL context is wrong in method parameter resolution [SPR-6525]

[spring-projects-issues]

Dave Syer opened SPR-6525 and commented

SpEL context is wrong in method parameter resolution. Instead of using the current context for the stack (which refers to the method target), parameters should have a current context equal to the context root. This test would fail in EvaluationTests():

@Test
public void testNestedMethod() {
     evaluate("placeOfBirth.doubleIt(inventions.length)", 18, Integer.class);
}

---

Affects: 3.0 RC3

Attachments:

• SPR-6526.patch (1.51 kB)

Issue Links:

• Spel parser is not able to resolve bean id variable when it is passed as parameter to a static method. [SPR-5584] #10255 Spel parser is not able to resolve bean id variable when it is passed as parameter to a static method. ("is duplicated by")

Referenced from: commits 33a7bbb

[spring-projects-issues]

Dave Syer commented

Attached #11191.patch. I'll leave it up to Andy to decide if it has any side effects (the tests all pass).

[spring-projects-issues]

Andy Clement commented

I posted on spr-dev but that is waiting moderation.

Mark F mentioned this at spring one and I meant to look at it but ran out of time. The workaround is to qualify the reference in the parameter:

placeOfBirth.doubleIt(#root.inventions.length)

The behaviour should change as you suggest, but I'm inclined (because of the trivial workaround - if that does work!) to put it into 3.0.1 and 3.1.

[spring-projects-issues]

Sam Brannen commented

In any case, please add a section to the reference manual covering "#root" in contrast to "#this".

Thanks,

Sam

[spring-projects-issues]

Dave Syer commented

Yeuch. Ugly workaround for a really common use case. I think we should fix it, otherwise we are going to get a lot of support requests. Can't we just add the patch quickly and fix anything that crops up as a result in 3.0.1?

The workaround does work, by the way.

[spring-projects-issues]

Andy Clement commented

personally I prefer an RC to be an RC - so this would necessitate an RC4. but if everyone else thinks it should go in regardless and not require another RC I can be persuaded.

[spring-projects-issues]

Andy Clement commented

Probably also affects constructor invocation and function invocation

[spring-projects-issues]

Andy Clement commented

per Juergens comment, I'll look at this tomorrow. I will look at constructors and function invocation too.

[spring-projects-issues]

Andy Clement commented

test and fix (per Dave's patch) is in. Added extra tests for function invocation, ctor invocation and a combined method/function invocation.

Also added the, err, 'secret' #root to the docs.

[spring-projects-issues]

Stevo Slavić commented

#10255 duplicates this issue.