Remove injection and lookups for raw maps according to challenge #1710

[arjantijms]

Fixes #1710

Signed-off-by: Arjan Tijms [email protected]

[brideck]

This PR should also remove/disable the corresponding tests:

• faces23 - Spec1323IT (ApplicationMap)
• faces23 - Spec1335IT (ViewMap)
• faces23 - Spec1345IT (SessionMap)
• faces23 - Spec1353IT (RequestCookie)
• faces40 - Spec1582ApplicationMapIT
• faces40 - Spec1582RequestCookieMapIT
• faces40 - Spec1582ViewMapIT

[brideck]

It would appear that I was hasty with my initial review. Talked this over with some of our devs and the remaining faces40/cdi tests would need an update, too.

In the beans, lines like:

Map<String, String> headerMap = CDI.current().select(Map.class, HeaderMap.Literal.INSTANCE).get();

would need to be updated to:

Map<String, String> headerMap = CDI.current().select(new TypeLiteral<Map<String, String>>() {}, HeaderMap.Literal.INSTANCE).get();

in order to correctly test what's being attempted here.

[brideck]

It's also worth pointing out that the ApplicationMap test (both faces23 and faces40) are verifying behavior using the Mojarra-specific property jakarta.faces.BEANS_VALIDATION_AVAILABLE, which won't work for us. Can that be fixed here, too?

Alternatively, I could add it to the challenge for implementation-specific grievances (#1703). Or if it's preferred, I can open a new challenge for additional implementation-specific grievances.

[arjantijms]

Map<String, String> headerMap = CDI.current().select(new TypeLiteral<Map<String, String>>() {}, HeaderMap.Literal.INSTANCE).get();

In a way for usability this is quite a step back, but a logical consequence of not allowing raw maps indeed. Maybe we can introduce something like a HeaderMap.TYPE to represent new TypeLiteral<Map<String, String>>() {} for Faces.Next.

[brideck]

This same change needs to be made to the faces23/cdi version of this test, too.