teh

Points 1 and 2 sound like a good idea! I had forgotten that we're storing the `GType` in VariableDefinitions. Need a bit more time to understand part 3.

@jamesdabbs-procore you are right, recursion can't be expressed directly with `type`. You can work around that with an opaque (to the type system) `newtype`. E.g. from one my my projects:...

@jml Not super sure why the coverage ratchet passed for 8.0 but not for 8.2. Do you have any insight / ideas?

You are right @theobat - the only way to construct a result is via a handler. The idea at the time IIRC was that we only wanted to execute a...

Thanks for the pointer! I think I'd like to maybe wait out one more release while we work out the roughest edges. E.g. today I realized that our tests don't...

@jasonzoladz please raise specific issues for changes or fixes. I'm currently implementing a project using this library, and I'm making a list of what I want changed. The more use...

Code in question: https://github.com/jml/graphql-api/blob/master/src/GraphQL/Resolver.hs#L433-L444 One solution would be to relax the constraint for what types we allow in unions. We can resolve anything that has a `HasResolver` instance, but OTOH...

Hi - this library doesn't support introspection, so GraphiQL might work for running queries, but won't work for type completion, docs etc. Does that make sense?

I didn't turn it out explicitly but I can reproduce how it got turned on: by depending on an extension that in turn depends on a `prebuilt_cxx_library`: ```bash mkdir hello...

I can confirm that stops building libomnibus, thank you! I think the original bug (don't rewrite weak symbols to undefined symbols) is still valid so will leave closing of issue...