Fernando Jiménez Moreno [:ferjm] Reporter
	Description • 10 years ago

I have the following WebIDL:

[...]

[NavigatorProperty="mozPaymentProvider",
 NoInterfaceObject,
 HeaderFile="mozilla/dom/PaymentProviderUtils.h",
 Func="mozilla::dom::PaymentProviderUtils::EnabledForScope",
 JSImplementation="@mozilla.org/payment/provider;1"]
interface PaymentProvider {

[...]

  [Cached, Pure, Frozen] readonly attribute sequence<PaymentIccInfo> iccInfo;

and I am getting this build error:

/builds/slave/try-l64-0000000000000000000000/build/src/obj-firefox/dom/bindings/MozPaymentProviderBinding.cpp:773:1: error: 'bool mozilla::dom::PaymentProviderBinding::_ClearCachedIccInfoValue(JSContext*, unsigned int, JS::Value*)' defined but not used [-Werror=unused-function]

I couldn't find these combination of attributes in any WebIDL JS implemented, so I was wondering if this is a DOM bindings issue.

	Fernando Jiménez Moreno [:ferjm] Reporter
	Comment 1 • 10 years ago

This is the generated dom/bindings/MozPaymentProviderBinding.cpp

	Boris Zbarsky [:bzbarsky] Assignee
	Comment 2 • 10 years ago

Yep.  The method is output based on this condition:

        if descriptor.interface.isJSImplemented():

but used based on this condition:

        if (descriptor.interface.isJSImplemented() and
            descriptor.interface.hasInterfaceObject()):

	Boris Zbarsky [:bzbarsky] Assignee
	Comment 3 • 10 years ago

Seems to me like the latter is actually wrong.

	Andrew McCreight [:mccr8]
	Comment 5 • 10 years ago

Comment on attachment 8543379 [details] [diff] [review]
Output the JS hooks for our cache-clearing stuff for [Pure] properties on JS-implemented objects even if we have no interface object, since these live on instances

Review of attachment 8543379 [details] [diff] [review]:
-----------------------------------------------------------------

Makes sense.  Thanks for fixing this!

	Andrew McCreight [:mccr8]
	Comment 6 • 10 years ago

Oh, and it might be a good idea to add a case to the test WebIDL file for this.  Though soon enough I guess we'll have something in the tree that triggers it.

	Boris Zbarsky [:bzbarsky] Assignee
	Comment 7 • 10 years ago

Yeah, we'd have to add a whole new interface to the test IDL to test this... I guess I could do that.

	Boris Zbarsky [:bzbarsky] Assignee
	Comment 8 • 10 years ago

Did that, and https://hg.mozilla.org/integration/mozilla-inbound/rev/b12bf3c3d05a

	Phil Ringnalda (:philor)
	Comment 9 • 10 years ago

https://hg.mozilla.org/mozilla-central/rev/b12bf3c3d05a