What shoud script element's textContent's sink string be?

[mbrodesser-Igalia]

https://w3c.github.io/trusted-types/dist/spec/#dom-htmlscriptelement-textcontent states "HTMLScriptElement textContent".

A WPT requires "HTMLScriptElement text":
https://searchfox.org/mozilla-central/rev/1f2c929f93eabe6d753096e47641ebba68b16601/testing/web-platform/tests/trusted-types/default-policy-callback-arguments.html#27-28,45-46

[fred-wang]

https://w3c.github.io/trusted-types/dist/spec/#dom-htmlscriptelement-textcontent states "HTMLScriptElement textContent".

I think the spec is quite clear, it should be "HTMLScriptElement textContent".

A WPT requires "HTMLScriptElement text": https://searchfox.org/mozilla-central/rev/1f2c929f93eabe6d753096e47641ebba68b16601/testing/web-platform/tests/trusted-types/default-policy-callback-arguments.html#27-28,45-46

This test checks the correct sink names here:

assert_equals(args[2], current_case[2], "Expecting the sink name.");

The switch statement you mention is a noop: args[1] is supposed to be the expectedType’s type (see https://w3c.github.io/trusted-types/dist/spec/#process-value-with-a-default-policy-algorithm and https://w3c.github.io/trusted-types/dist/spec/#abstract-opdef-get-trusted-type-policy-value) which is a string while TrustedHTML, TrustedScript and TrustedScriptURL are objects.

The same mistake seems to happen for the switch statement added in web-platform-tests/wpt#46089

I'll try and fix that.

cc @ziransun @lukewarlow

[fred-wang]

This is fixed by web-platform-tests/wpt#49909

(I cannot close the issue because my github account does not have proper permission for this repo)

[evilpie]

I think the spec is quite clear, it should be "HTMLScriptElement textContent".

To me it looks like the test that was updated by web-platform-tests/wpt#49909 still uses "HTMLScriptElement text"

https://github.com/web-platform-tests/wpt/blob/9a0b9963e02f06989a4f63a80b7091956a1d23c9/trusted-types/default-policy-report-only.html#L77

[lukewarlow]

That test should be text because that's the sink that's tested in it.

[fred-wang]

Yeah basically when setting HTMLScriptElement's textContent, the sink string should be textContent. And when setting HTMLScriptElement's text the sink string should be text.

The confusion was with default-policy-callback-arguments.html, which was setting HTMLScriptElement's textContent but had two contradictory assertions for the sink string (checking it's both textContent and text).

As explained above, the second assertion was actually not reachable (and wrong) so I removed it. I took the opportunity to fix unreachable assertions in other tests.

[evilpie]

Sorry, makes sense!

[lukewarlow]

Great catch btw sorry I didn't spot that mistake when reviewing/writing the relevant tests.