Schema#multiplex causes context.namespace(:interpreter)[:current_path] to be nil inside `then` block

[DmitryTsepelev]

Describe the bug

I found a case when context.namespace(:interpreter)[:current_path] is nil (to be precise, most of the keys inside the :interpreter hash are missing). The problem appears when Schema#multiplex is called and I try to get the current path inside .then block in promise, i.e.:

class Tweet < GraphQL::Schema::Object
    field :content, String, null: false
    field :author, Types::User, null: false

    def author
      puts "Tweet:author current_path=#{context.namespace(:interpreter)[:current_path]}"

      Batch::RecordLoader.for(::User).load(object.author_id).then do |author|
        puts "Tweet:author after batch current_path=#{context.namespace(:interpreter)[:current_path]}"
        author
      end
    end
  end

However, it works fine when Schema#execute is used.

Not sure if I should go to the graphql-batch repo with this issue.

Versions

graphql version: latest graphql-batch version: latest

Steps to reproduce

A whole example:

```ruby begin require "bundler/inline" rescue LoadError => e $stderr.puts "Bundler version 1.10 or later is required. Please update your Bundler" raise e end gemfile(true) do source "https://rubygems.org" gem "rails", "6.0.3" gem "pg" gem "graphql", "~> 1.12" gem "rspec-rails", "~> 4.0.1" gem "db-query-matchers" gem "ar_lazy_preload" gem "graphql-batch" end require "active_record" class App < Rails::Application config.logger = Logger.new('/dev/null') end App.initialize! ActiveRecord::Base.establish_connection(adapter: "postgresql", database: "nplusonedb") ActiveRecord::Schema.define do enable_extension "plpgsql" create_table "users", force: :cascade do |t| t.string "nickname", null: false t.datetime "created_at", precision: 6, null: false t.datetime "updated_at", precision: 6, null: false end create_table "tweets", force: :cascade do |t| t.text "content", null: false t.bigint "author_id", null: false t.datetime "created_at", precision: 6, null: false t.datetime "updated_at", precision: 6, null: false end add_foreign_key "tweets", "users", column: "author_id" end class User < ActiveRecord::Base has_many :tweets, foreign_key: :author_id end class Tweet < ActiveRecord::Base belongs_to :author, class_name: "User" end module Batch class RecordLoader < GraphQL::Batch::Loader def initialize(model) @model = model end def perform(ids) @model.where(id: ids).each { |record| fulfill(record.id, record) } ids.each { |id| fulfill(id, nil) unless fulfilled?(id) } end end end module Types class User < GraphQL::Schema::Object field :nickname, String, null: false end class Tweet < GraphQL::Schema::Object field :content, String, null: false field :author, Types::User, null: false def author puts "Tweet:author current_path=#{context.namespace(:interpreter)[:current_path]}" Batch::RecordLoader.for(::User).load(object.author_id).then do |author| puts "Tweet:author after batch current_path=#{context.namespace(:interpreter)[:current_path]}" author end end end end class FeedResolver < GraphQL::Schema::Resolver type [Types::Tweet], null: false def resolve(limit: nil, offset: nil) Tweet.limit(limit).offset(offset) end end module Types class Query < GraphQL::Schema::Object field :feed, [Types::Tweet], null: false, resolver: FeedResolver do argument :limit, Integer, required: false argument :offset, Integer, required: false end end end class GraphqlSchema < GraphQL::Schema query Types::Query use GraphQL::Batch end Tweet.delete_all User.delete_all john = User.create(nickname: "John") max = User.create(nickname: "Max") john.tweets.create(content: "Hi!", created_at: Time.new(2020, 6, 12, 10)) max.tweets.create(content: "Hello!", created_at: Time.new(2020, 6, 13, 10)) john.tweets.create(content: "My second tweet is here", created_at: Time.new(2020, 6, 15, 10)) max.tweets.create(content: "The weather is nice", created_at: Time.new(2020, 6, 17, 10)) puts "\nSchema#multiplex\n" query1 = <<~GQL query { feed(offset: 0, limit: 2) { content author { nickname } } } GQL query2 = <<~GQL query { feed(offset: 2, limit: 2) { content author { nickname } } } GQL puts GraphqlSchema.multiplex([{ query: query1 }, { query: query2 }]).map(&:to_h).inspect puts "\nSchema#execute\n" puts GraphqlSchema.execute(query1).to_h.inspect ```

Expected behavior

I want to be able to get the current path inside the promise 🙂

Actual behavior

Current path is available only inside first promise block. Here is what I see in the console:

Schema#multiplex
Tweet:author current_path=["feed", 0, "author"]
Tweet:author current_path=["feed", 1, "author"]
Tweet:author current_path=["feed", 0, "author"]
Tweet:author current_path=["feed", 1, "author"]
Tweet:author after batch current_path=["feed", 0, "author"]
Tweet:author after batch current_path=
Tweet:author after batch current_path=["feed", 0, "author"]
Tweet:author after batch current_path=
[{"data"=>{"feed"=>[{"content"=>"Hi!", "author"=>{"nickname"=>"John"}}, {"content"=>"Hello!", "author"=>{"nickname"=>"Max"}}]}}, {"data"=>{"feed"=>[{"content"=>"My second tweet is here", "author"=>{"nickname"=>"John"}}, {"content"=>"The weather is nice", "author"=>{"nickname"=>"Max"}}]}}]

Schema#execute
Tweet:author current_path=["feed", 0, "author"]
Tweet:author current_path=["feed", 1, "author"]
Tweet:author after batch current_path=["feed", 0, "author"]
Tweet:author after batch current_path=["feed", 0, "author"]

[rmosolgo]

Hey, thanks for the details on this issue! Yes, this sounds like a bug to me, and this is the right place to report it.

I would have expected those context values to be populated by this:

https://github.com/rmosolgo/graphql-ruby/blob/e1516208593ce15c059c4066a8b045e5e5cf9f1b/lib/graphql/execution/interpreter/runtime.rb#L489

But evidently that's not the case!

[rmosolgo]

(derp, wrong button.)

[DmitryTsepelev]

I've tried to add logging right after graphql-ruby/lib/graphql/execution/interpreter/runtime.rb:489 and figured out, that it's called once before resolving the lazy object (and 3 more times after):

Schema#multiplex
Tweet:author current_path=["feed", 0, "author"]
Tweet:author current_path=["feed", 1, "author"]
Tweet:author current_path=["feed", 0, "author"]
Tweet:author current_path=["feed", 1, "author"]
set_all_interpreter_context(owner_object, #<GraphQL::Schema::Field:0x00007fcf0af871f0>, {}, ["feed", 0, "author"])
Tweet:author after batch current_path=["feed", 0, "author"]
Tweet:author after batch current_path=
Tweet:author after batch current_path=["feed", 0, "author"]
Tweet:author after batch current_path=
set_all_interpreter_context(owner_object, #<GraphQL::Schema::Field:0x00007fcf0af871f0>, {}, ["feed", 1, "author"])
set_all_interpreter_context(owner_object, #<GraphQL::Schema::Field:0x00007fcf0af871f0>, {}, ["feed", 0, "author"])
set_all_interpreter_context(owner_object, #<GraphQL::Schema::Field:0x00007fcf0af871f0>, {}, ["feed", 1, "author"])
[{"data"=>{"feed"=>[{"content"=>"Hi!", "author"=>{"nickname"=>"John"}}, {"content"=>"Hello!", "author"=>{"nickname"=>"Max"}}]}}, {"data"=>{"feed"=>[{"content"=>"My second tweet is here", "author"=>{"nickname"=>"John"}}, {"content"=>"The weather is nice", "author"=>{"nickname"=>"Max"}}]}}]

[rmosolgo]

Ohhh, interesting. I guess it's because the promise created by .then is actually executed at graphql-batch's discretion. It probably runs that promise while running the one for the previous field, which would be before the Execution::Lazy which contains the promise is executed. So, GraphQL-Ruby has no chance of updating context in that case -- control flow is in the hands of graphql-batch instead.

As a work-around, you might have to capture context[:current_path] outside the promise and somehow pass it in (using lexical scope) to the code inside it.

Otherwise, I'm not sure how GraphQL-Ruby could inject runtime info (for other fields) into GraphQL-batch's promise resolution 😖

Although practically irrelevant, it looks like GraphQL::Dataloader "just works" in this regard. Thanks for your excellent reproduction of this issue -- when I modified it to use GraphQL::Dataloader, it outputted the expected:

Diff of replication example

```diff git diff --no-index original_issue.rb modified_issue.rb diff --git a/original_issue.rb b/modified_issue.rb index 9cf7ec80a..36f46bbde 100644 --- a/original_issue.rb +++ b/modified_issue.rb @@ -54,14 +54,15 @@ class Tweet < ActiveRecord::Base end module Batch - class RecordLoader < GraphQL::Batch::Loader - def initialize(model) - @model = model + class RecordSource < GraphQL::Dataloader::Source + def initialize(model_class) + @model_class = model_class end - def perform(ids) - @model.where(id: ids).each { |record| fulfill(record.id, record) } - ids.each { |id| fulfill(id, nil) unless fulfilled?(id) } + def fetch(ids) + records = @model_class.where(id: ids) + # return a list with `nil` for any ID that wasn't found + ids.map { |id| records.find { |r| r.id == id.to_i } } end end end @@ -78,10 +79,9 @@ module Types def author puts "Tweet:author current_path=#{context.namespace(:interpreter)[:current_path]}" - Batch::RecordLoader.for(::User).load(object.author_id).then do |author| - puts "Tweet:author after batch current_path=#{context.namespace(:interpreter)[:current_path]}" - author - end + author = dataloader.with(Batch::RecordSource, ::User).load(object.author_id) + puts "Tweet:author after batch current_path=#{context.namespace(:interpreter)[:current_path]}" + author end end end @@ -105,7 +105,7 @@ end class GraphqlSchema < GraphQL::Schema query Types::Query - use GraphQL::Batch + use GraphQL::Dataloader end Tweet.delete_all ```

[DmitryTsepelev]

Yeah, I don't see any clear way to update the context inside the promise block. Closing the issue, thank you for the help!