Olivier Mélois

> Or does prism/highlight/whatever support smithy now? ~It does :) it's called "linguist"~ eeeeeergh yeah you're right, we're using prism on the microsite -_____-'

Apparently docusaurus can use the shiki highlighter, which uses textmate grammars, which is what we use to define highlighting in vscode. https://www.npmjs.com/package/docusaurus-preset-shiki-twoslash So I believe if we found a way...

And here are the docs on how to do just that : https://github.com/shikijs/shiki/blob/main/docs/languages.md#supporting-your-own-languages-with-shiki Now it'd just be a matter of understanding how to wire this altogether

> ok , so IIUC , we can keep the smithy annotation and benefit from linguist support for Smithy, we would just need to provide prism with a custom language...

Hi @caoilte, in the current stateit'd be possible to write something like : ```  trait Validator[A] { def validate(a: A) : Either[List[String], A] } object ValidatorSchemaVisitor extends SchemaVisitor[Validator] { ......

> would all types with refinements get different constructors, or just newtypes? probs just newtypes. > would the new constructor be in addition to the current one, or instead of?...

I don't have bandwidth to offer guidance on this, besides maybe looking at previous art: https://github.com/monix/newtypes/blob/main/core/shared/src/main/scala/monix/newtypes/NewValidated.scala

Not characterising it as a bug, but it is an unfortunate consequence of the combinations of two features that are working well individually. I think we're gonna have to target...

Well I'm not sure whether your reasoning makes sense as a justification to not address the problem. As a matter of fact, it reinforces the justification to fix it, as...

Yeah, I'm thinking we should get rid of the `UnknownErrorResponse` in favour of something like : ``` RawErrorResponse(status, headers, body, failedDecodeAttempt: Option[FailedDecodeAttempt]) FailedDecodeAttempt(discriminator, contractError) ```