[OAS 3.1.0] New restrictions on ServerVariable.enum

[Azquelt]

Two new restrictions are added to ServerVariable.enum:

• If enum is defined it may not be empty
• If enum is defined then the value set for default must be found within enum

We should

• Check that we don't have any tests which violate this constraint
• Update our documentation if necessary

[benjamin-confino]

tck/src/main/java/org/eclipse/microprofile/openapi/tck/ModelConstructionTest.java

This test modifies the enum list, but never sets a default so the ServerVariable will be in an illegal state.

(It also raises the question of intermediate values. If you want to remove the defualt from the enum, add a new value, snd set the new value to the default. Is it required to ensure the default is in the enum at every step of the process?)

The following tests use ServerVariable in line with the rule

tck/src/main/java/org/eclipse/microprofile/openapi/apps/airlines/JAXRSApp.java - enumeration is unset.

tck/src/main/java/org/eclipse/microprofile/openapi/apps/airlines/resources/ReviewResource.java - enumeration is set, default is set and in the enumeration.

tck/src/main/java/org/eclipse/microprofile/openapi/reader/MyOASModelReaderImpl.java - enumeration is set, default is set and in the enumeration.

The following files only mention ServerVariable

api/src/main/java/org/eclipse/microprofile/openapi/models/servers/Server.java
api/src/main/java/org/eclipse/microprofile/openapi/models/servers/ServerVariable.java
api/src/main/java/org/eclipse/microprofile/openapi/models/servers/package-info.java
api/src/main/java/org/eclipse/microprofile/openapi/OASFactory.java
api/src/main/java/org/eclipse/microprofile/openapi/annotations/servers/Server.java
api/src/main/java/org/eclipse/microprofile/openapi/annotations/servers/ServerVariable.java
api/src/main/java/org/eclipse/microprofile/openapi/annotations/servers/package-info.java

[Azquelt]

(It also raises the question of intermediate values. If you want to remove the defualt from the enum, add a new value, snd set the new value to the default. Is it required to ensure the default is in the enum at every step of the process?)

No, I don't think this is required, the object being in an invalid intermediate state is fine.

[benjamin-confino]

I spoke with Andrew yesterday and learned that we do not add TCK tests for illegal states. Given that, and the results of my audit above, I think this issue only needs an update to the documentation.

[Azquelt]

I think we should doc that default must be one of the values in enum if enum is set. Do this on the default field in the ServerVariable model interface and annotation.