Bug 24523 – Add test to check that an invalid second argument of window.postMessage is ignored

• Status: RESOLVED
• Resolution: FIXED
• Component: DOM

[Pam Greene (IRC:pamg)]

Make sure that the second argument of window.postMessage is ignored if it's not a message port.

[Pam Greene (IRC:pamg)]

Created attachment 28495 [details]
New test + result

[Alexey Proskuryakov]

Is this correct behavior, or an implementation quirk? I doubt that I fully investigated this case when writing the code - but HTML5 seems to imply that an exception should be raised. What does Firefox do?

I don't think that using the test framework here helps much, as we don't have a long series of tests, and this isn't a fast/js test. As mentioned in another bug, such usage of successfullyParsed won't work in most interesting cases.

[Pam Greene (IRC:pamg)]

Created attachment 28534 [details]
Patch addressing Alexey's comments

(In reply to comment #2)
> Is this correct behavior, or an implementation quirk? I doubt that I fully
> investigated this case when writing the code - but HTML5 seems to imply that an
> exception should be raised. What does Firefox do?

Firefox (Mac 3.0.7) raises an exception.  The HTML4 spec doesn't say what to do. HTML5 does imply that an exception should be raised. I'd suggest adding this test as it is for now, to catch unintentional changes, and updating it when there's a deliberate switchover to HTML5 behavior.

> I don't think that using the test framework here helps much, as we don't have a
> long series of tests, and this isn't a fast/js test. As mentioned in another
> bug, such usage of successfullyParsed won't work in most interesting cases.

Yes, notice that successfullyParsed is not used in the normal way here, just inserted to make the output look like what people expect to see.

In any case, I've pulled out the test framework in this patch, since as you say it's not adding much.

[Alexey Proskuryakov]

Comment on attachment 28534 [details]
Patch addressing Alexey's comments

r=me - at least this tests that we don't crash!

Please add a comment saying that the results are likely incorrect, so that the test doesn't scare away the person fixing the behavior.

[Pam Greene (IRC:pamg)]

Comment added. Landed in r41651.