Bug 189283 – [WebAuthN] Polish WebAuthN auto-test environment

• Status: RESOLVED
• Resolution: FIXED
• Component: WebCore Misc.

[Jiewen Tan]

Currently, tests of WebAuthN spread in three places:
1) LayoutTests/http/tests/webauthn/
2) LayoutTests/http/tests/wpt/webauthn/
3) TestWebKitAPI/Tests/ios/LocalAuthenticator.mm

It will be great to have them all in LayoutTests/http/tests/wpt/webauthn/. This could be achieved by:
1) making tests under LayoutTests/http/tests/webauthn/ become wpt async tests.
2) mocking transport layer abstraction, says HIDService, and providing necessary test configuration functions in testRunner.idl such that layout tests could then directly manipulate the most underlying layer of WebAuthN to have test behaviors/results they expected.

[Radar WebKit Bug Importer]

<rdar://problem/44117828>

[Jiewen Tan]

Created attachment 351016 [details]
Patch

[Jiewen Tan]

Created attachment 351019 [details]
Patch

[Chris Dumez]

Comment on attachment 351019 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=351019&action=review

r=me

> Source/WebKit/UIProcess/API/C/WKWebsiteDataStoreRef.cpp:582
> +        MockWebAuthenticationConfiguration::Local local;

Any reason we are using a local variable instead of assigning straight into configuration.local?

> Source/WebKit/UIProcess/WebAuthentication/Mock/MockLocalConnection.mm:77
> +        if (!configuration.local.value().acceptAuthentication) {

Cannot we use onfiguration.local->acceptAuthentication ?

> Source/WebKit/UIProcess/WebAuthentication/Mock/MockLocalConnection.mm:90
> +        if (!configuration.local.value().acceptAuthentication) {

ditto.

> Source/WebKit/UIProcess/WebAuthentication/Mock/MockLocalConnection.mm:105
> +        if (!configuration.local.value().acceptAttestation) {

ditto.

> Source/WebKit/UIProcess/WebAuthentication/Mock/MockLocalConnection.mm:118
> +            (__bridge CFDataRef)adoptNS([[NSData alloc] initWithBase64EncodedString:configuration.local.value().privateKeyBase64 options:NSDataBase64DecodingIgnoreUnknownCharacters]).get(),

ditto.

> Source/WebKit/UIProcess/WebAuthentication/Mock/MockLocalService.cpp:45
> +    if (!m_configuration.local)

return !!m_configuration.local; ?

[Jiewen Tan]

Comment on attachment 351019 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=351019&action=review

Thanks Chris for r+ my patch.

>> Source/WebKit/UIProcess/API/C/WKWebsiteDataStoreRef.cpp:582
>> +        MockWebAuthenticationConfiguration::Local local;
> 
> Any reason we are using a local variable instead of assigning straight into configuration.local?

configuration.local is a std::optional. How?

>> Source/WebKit/UIProcess/WebAuthentication/Mock/MockLocalConnection.mm:77
>> +        if (!configuration.local.value().acceptAuthentication) {
> 
> Cannot we use onfiguration.local->acceptAuthentication ?

Sure.

>> Source/WebKit/UIProcess/WebAuthentication/Mock/MockLocalConnection.mm:90
>> +        if (!configuration.local.value().acceptAuthentication) {
> 
> ditto.

Fixed.

>> Source/WebKit/UIProcess/WebAuthentication/Mock/MockLocalConnection.mm:105
>> +        if (!configuration.local.value().acceptAttestation) {
> 
> ditto.

Fixed.

>> Source/WebKit/UIProcess/WebAuthentication/Mock/MockLocalConnection.mm:118
>> +            (__bridge CFDataRef)adoptNS([[NSData alloc] initWithBase64EncodedString:configuration.local.value().privateKeyBase64 options:NSDataBase64DecodingIgnoreUnknownCharacters]).get(),
> 
> ditto.

Fixed.

>> Source/WebKit/UIProcess/WebAuthentication/Mock/MockLocalService.cpp:45
>> +    if (!m_configuration.local)
> 
> return !!m_configuration.local; ?

I don't think so.

[Chris Dumez]

Comment on attachment 351019 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=351019&action=review

>>> Source/WebKit/UIProcess/API/C/WKWebsiteDataStoreRef.cpp:582
>>> +        MockWebAuthenticationConfiguration::Local local;
>> 
>> Any reason we are using a local variable instead of assigning straight into configuration.local?
> 
> configuration.local is a std::optional. How?

Ah, I missed that it was an optional. Please disregard this comment.

>>> Source/WebKit/UIProcess/WebAuthentication/Mock/MockLocalService.cpp:45
>>> +    if (!m_configuration.local)
>> 
>> return !!m_configuration.local; ?
> 
> I don't think so.

Why not? :)

[Jiewen Tan]

Created attachment 351134 [details]
Patch for landing

[Jiewen Tan]

Comment on attachment 351019 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=351019&action=review

>>>> Source/WebKit/UIProcess/WebAuthentication/Mock/MockLocalService.cpp:45
>>>> +    if (!m_configuration.local)
>>> 
>>> return !!m_configuration.local; ?
>> 
>> I don't think so.
> 
> Why not? :)

I was thinking you suggested me to replace !m_configuration.local with !!m_configuration.local. lol.
Yup. Why not.

[Jiewen Tan]

Created attachment 351137 [details]
Patch for landing

[WebKit Commit Bot]

Comment on attachment 351137 [details]
Patch for landing

Clearing flags on attachment: 351137

Committed r236625: <https://trac.webkit.org/changeset/236625>