Description
I love this lib - thanks so much for your hard work, @syrusakbary!
With that said, I'd happily contribute to paying for more of your time to extend Graphene, and fill in a few missing pieces that would make it more viable to use in production.
Namely:
-
Error catching, per Solve dependency conflict between Flake8 and Sphinx graphql-core#177, and subscribe breaking change in graphql version 3.3.0a3 graphql-core#202. System exceptions bubble up to user output by default, which a dangerous default IMO. I'd love to see a cleaner way of logging exceptions and controlling error output that could be defined in one place, rather than defensive
try/exceptblocks in every mutation. -
Core improvements / re-factoring. Thread safety (Django relations not traversed #43), resolving promise issues (Assertion errors from promise when using graphene-django. syrusakbary/promise#57), moving to
async/awaitper your comment, etc. There are a few core things that are beyond my immediate experience with how things work under the hood, that seem like they have some potential to throw weird/unexpected issues that are hard to diagnose. -
Documented subscription support/patterns, per Does graphene support subscription now? #781.
-
Docs synced with releases. There have been a few occasions where trying doc examples has thrown errors, such as Middleware example fails in doc #812 today or passing context to GraphQLView no more supported? flask-graphql#52 which I ran into a few days ago. When this happens, monkey-patched workarounds are usually suggested by the community, which makes for brittler code.
-
Some more tooling to provide official solutions for Offer a way to calculate and limit the total cost of a query #772, etc. Calculating the cost of queries, parsing the AST tree and preempting query joins that might be necessary, etc. Outside the purview of the core, perhaps, but necessary stuff at scale that would save devs reinventing the wheel.
-
Closing issues faster. This is just a general thing, but there are issues going back 2-3 years that would be good to get unstuck. Much of it might even be redundant now or have other solutions, but clearing the backlog would make a clearer case for using the lib in production, knowing those same issues are unlikely to resurface.
Obviously, your time is valuable and the fact that you've put together anything at all - let alone something as cool as Graphene - is amazing, so thank you.
But perhaps there's a way, as a community, we could buy more of your time to address some of the above? I'd happily chuck in a few bucks on the reg to keep this lib up-to-date.