Anyone already thought about the new GraphQLContex...

# graphql-kotlin

[Joe]

Anyone already thought about the new GraphQLContext map and type-safe context storage? We currently have a context object with some well defined/typed properties; moving to a string key + a cast of the returned value feels like a step back even though it'd probably be fine. Possibly extensions to GraphQLContext for these well defined properties might make sense, but haven't fully thought this through yet, so wondering if others had?

[Dariusz Kuc]

fyi graphql-java relies on casting -> https://github.com/graphql-java/graphql-java/blob/master/src/main/java/graphql/ParseAndValidate.java#L50

[Shane Myrick]

The “cast” is still happening on our end in the old object. We just take the context and send it to the argument that we hope will match so in a way that is less safe.

You also don’t have to store in the map by string, you can store by anything that can be used as a key, even another object

For example, the graphql-java engine now uses this method internally to pass around config options for the parser

https://github.com/graphql-java/graphql-java/blob/992d95fe43e7b86a9299921d2ac1822bd3e9e204/src/main/java/graphql/ParseAndValidate.java#L50

[Dariusz Kuc]

lol timing

[Shane Myrick]

🦥 ⌨️

[Joe]

thanks, getting away from string keys does help a bit actually.

[Joe]

dealing with mapOf() when creating the context but dealing with GraphQLContext when consuming it is a little confusing, but that looks to be a graphql-java issue, not a graphql-kotlin one