Thierry Joyal

> if we want to be more cautious about breaking changes, then we could have a deprecation warning in Liquid::Drop#liquid_method_missing if @context is nil 👍 Since it is very hard...

Having some discussions, it looks like a possible avenue is to use a thread local instead of trying to manually set the `context=`. If there is a desire to keep...

> I suppose the biggest blockers to this are the two breaking changes identified so far, particularly to test code. I am really not worried about it. Tests will require...

To re-iterate, I am not worried about implementation details as much as I am about __code simplicity__(removal of manual developer responsibility) and __output accuracy__. If less code and less tests...

I am also struggling with this issue and I think we should work a path forward for it. That being said while this implementation allow retrieval, I see it problematic...

@jhawthorn something you can pitch in? 😸 💚

I've been struggling myself with this issue and I think we should fine a path to fix it. Linking items, https://github.com/github/graphql-client/pull/249 has an attempt for it.

No I'm sorry, it was a puzzling but not creating in my cases blockers or anything other than "know limitation". We ended up rolling with the current behaviour.

This has not been answered in a while now. I've been digging in for something else and found some information related to this issues. https://github.com/ruby/psych/blob/df242e764492b602ced499e367ccf339cdb36699/lib/psych/scalar_scanner.rb#L8-L9 https://yaml.org/type/timestamp.html It looks like Psych...

> There's not much context around why only Octokit::ServerError was included, and not the default list from Faraday Looking back at the PR, it looks like this is simply my...