changed default CoroutineScope to not use GlobalScope to avoid potential cancellation issues and other pitfalls

[bsara]

From the Kotlin docs we read the following (emphasis added):

When a coroutine is launched in the CoroutineScope of another coroutine, it inherits its context via CoroutineScope.coroutineContext and the Job of the new coroutine becomes a child of the parent coroutine's job. When the parent coroutine is cancelled, all its children are recursively cancelled, too.

However, when GlobalScope is used to launch a coroutine, there is no parent for the job of the new coroutine. It is therefore not tied to the scope it was launched from and operates independently.

In order to avoid creating threads that aren't tied to any parent scopes that may exist, GlobalScope has been replaced in this PR with just a normal CoroutineScope that uses the Dispatchers.Default context when no scope is provided via the DataFetchingEnvironment context (I.E. this normal CoroutineScope is the default scope).

[rrva]

I see that even the old code allowed you to configure the coroutine scope by storing it in the context. How do you do this for all resolvers pre this change (5.7.1)? @oliemansm @bsara

[bsara]

@rrva I just used the SchemaParserOptions in my Spring Boot app and had a bean for the options:

typealias CoroutineContextSelector = () -> CoroutineContext

@Configuration
class GraphQLConfig {

  @Bean
  fun coroutineContextSelector(): CoroutineContextSelector = { Dispatchers.DEFAULT }

  @Bean
  fun schemaParserOptions(coroutineContextSelector: CoroutineContextSelectort) = SchemaParserOptions
    .newOptions()
    .coroutineContextProvider(CoroutineContextProvider { coroutineContextSelector() }
    .build()
}

I do the same thing even after updating to 6.0.0, but I don't call build() at the end. If you're not using Spring Boot, you could do the same thing, just make sure that the same options are used when you set everything up.