[Brainstorming] Survey questions

[martinbonnin]

Opening this issue because a question came up while discussing nullability that would be great to ask in a survey and I'm writing it down before completely forgetting about it 🙃

Questions would be:

• What client are you using?
  • urql, apollo, relay, etc...
  • no client, I use fetch() directly
• Are you using codegen?
• Are you handling partial data?

The intersection of fetch() + codegen + partial data could potentially be impacted by some of the semantic nullability solutions and it'd be grand to have some input to help decide on the tradeoff.

Bonus point if we can ask for contact info and reach out to those folks (with their agreement of course).

See #55

[saihaj]

Good idea! we discussed doing a survey in on of the board meetings and sharing with GraphQL Weekly. Something I think we should definitely run.

We don't even need a full survey, we can strategize and do a X/Linkedin poll

[martinbonnin]

Another interesting question as we look into updating the GraphQL spec is how long would it take for folks to upgrade to the "new" way of doing things. Do they have a deprecation policy for their API? What is it?

[benjie]

Quick thoughts on some questions which would be useful to us (note many of these will need guidance notes!)

• how do you handle versioning?
  • no versioning
  • major release on a dedicated schedule
  • infrequent major release as and when necessary
  • no need for backwards compatibility - everything ships at once
• do you use application/graphql-response+json?
• do you make things nullable to allow for errors?
• does your client use a normalized cache?
• do you make use of partial responses
• do you control all of your clients (i.e. can you update them all when you release a new version of your API)?
• do you use static documents?
• do you use trusted documents? 😁
• where do you use GraphQL
  • over HTTP to web browsers
  • over HTTP to mobile/desktop apps
  • service-to-service communications
  • like an ORM
  • without a transport
  • other
• do you use GraphQL for realtime?
  • yes, via websockets
  • yes, via server-sent events
  • yes, via some other technique (or technique unknown)
  • no
• if you use federation, how many subgraphs do you have?
• have you heard of "semantic nullability"
• how does your client handle errors in query operations?
  • discards entire response on error
  • ignores errors and just renders what it can
  • attempts to render errors locally
• are you aware of disabling null propagation
• do you use incremental delivery (@stream/@defer)?
  • yes, the pre-June 2023 version
  • yes, the post-June 2023 version
  • no
• what percentage of the fields in your schema are deprecated?
• how many GraphQL object types does your schema contain?
• what programming language do you use on the server
• what programming language do you use on the client
• do you use @oneOf
• do you use a linter?
• do you use a formatter?
• do you have GraphQL auto-complete in your code editor?
• do you use "fragment co-location"?
• do you use "fragment masking"?
• do you use "fragment arguments"?

I should probably stop.

[martinbonnin]

Who's using application/graphql-response+json?