AnnotationRule formatting uncertainties

[romtsn]

I have a following Retrofit interface:

interface UserService {

    fun updateMarketingConsent(
        @Path("userId") userId: String,
        @Path("context") context: String,
        @Body body: MarketingConsent
    ): Completable

    fun showUser(@Path("id") userId: String, @Query("include") include: String? = null, @QueryMap fields: Map<String, String> = emptyMap()): Single

    fun logout(@Path("userId") userId: String): Completable
}

Running ktlint in --format mode gives the following output:

interface UserService {

    fun updateMarketingConsent(
        @Path("userId")
        userId: String,
        @Path("context")
        context: String,
        @Body body: MarketingConsentStructure
    ): Completable

    fun showUser(
        @Path("id")
        userId: String,
        @Query("include") include: String? = null, @QueryMap fields: Map<String, String> = emptyMap()
    ): Single

    fun logout(
        @Path("userId")
        userId: String
    ): Completable
}

Imo, the latter is way less readable. I went through the official styleguide, but I think an annotation with arguments should be allowed for method params (even though not sure that it's applicable for all possible cases).

Also in the showUser function, the last two parameters should be placed on different lines (so the with-argument annotation followed by no-argument annotation). Cause otherwise the parameter-list-wrapping rules kicks in. This is most likely a bug in formatter, right?

[shashachu]

Ah, yeah. I don't think the rule was intended for parameter annotations. We can put up a diff to have it ignore those annotations completely.

[francescocervone]

Is there a way to disable the annotation rule?

[romtsn]

@francescocervone disabled_rules=experimental:annotation in your .editorconfig

[arturbosch]

+1 for this change

[yukukotani]

I think readability and following official style are both important. I suggest to create two codestyle (#548), official one and strong one.