Scott Triglia

Can you post your full traceback and precisely what line of code produces it? I think this isn't the first time working with semi-complete specs caused a low-level error....shouldn't spec...

Thanks for the fix Daine! This isn't the first bug we've had loading ini files....do you have any idea how we can effective test ini features? I'd love to write...

Thanks for asking Caitlin...you aren't the only one interested. The main blocker is support for marshalling and validation of OpenAPI 3 in the bravado-core library (see https://github.com/Yelp/bravado-core/issues/233). Once that is...

That seems...unpleasant. When you say "thread safety" you mean we're trying to access a URL-based ref before we've actually fetched it? Is it a racey issue or consistently reproducible? I'm...

Curious...the mere use of the `resolving` generator looks reasonable to me and I don't see where we are capable of ending up with nothing on the `scopes_stack` variable. Perhaps [pushing...

Ah! This seems likely to be quite relevant https://github.com/Julian/jsonschema/issues/223. Quoted: > It's not explicitly a goal to be threadsafe, generally I'd recommend just creating a different validator for each thread,...

Yeah we've been using it at decent scale (100qps+) at Yelp and I'm not aware of any issues here. We use uwsgi across different processes here, so we aren't exposed...

Ah! Then is the value of this ticket just "refactor function names to be clearer about which version they validate for"? I'm definitely not following the subtle point you raise....

Ah. Sounds like we should: 1. Add a test that omits a non200 status in the response and asserts we raise an error and 2. Fix the behavior. I think...

@ergo can you open up a separate issue on the subject of resource file paths? I think I'd find an example useful as I'm not entirely sure what you mean...