Is it possible to sort a union on a field that exists on all types in the union?

Users Dgraph
1

To facilitate better answering of questions, if you have a question, please fill in the following info. Otherwise, please delete the template.

What I want to do

I want to sort a union field by a field that exists on all types in the union. Pretty sure this isn’t possible, but I want to order the union field (Paragraph.text) not by just one field on one of union types (Word | Phrase), but by the .start field on both types. Phrase currently doesn’t have a .start field, was hoping I could use a var in place of the .start field on Phrase so that the value is calculated dynamically.

Schema

type Paragraph @auth(
  query: { rule: "{ $ROLE: { eq: \"ADMIN\" } }" }
  add: { rule: "{ $ROLE: { eq: \"ADMIN\" } }" }
  update: { rule: "{ $ROLE: { eq: \"ADMIN\" } }" }
  delete: { rule: "{ $ROLE: { eq: \"ADMIN\" } }" }
) {
  id: ID!
  start: Float
  end: Float
  speaker: Speaker!
  text: [Text]
  words: [Word!] @hasInverse(field: paragraph)
  phrases: [Phrase] @hasInverse(field: paragraph)
  transcript: Transcript!
}

type Phrase @auth(
  query: { rule: "{ $ROLE: { eq: \"ADMIN\" } }" }
  add: { rule: "{ $ROLE: { eq: \"ADMIN\" } }" }
  update: { rule: "{ $ROLE: { eq: \"ADMIN\" } }" }
  delete: { rule: "{ $ROLE: { eq: \"ADMIN\" } }" }
) {
  id: ID!
  paragraph: Paragraph!
  words: [Word!]! @hasInverse(field: phrase)
  start: Float # ideally should be a virtual field, phrase.words[0].start
  end: Float # ideally should be a virtual field, phrase.words[words.length - 1].end
}

type Word @auth(
  query: { rule: "{ $ROLE: { eq: \"ADMIN\" } }" }
  add: { rule: "{ $ROLE: { eq: \"ADMIN\" } }" }
  update: { rule: "{ $ROLE: { eq: \"ADMIN\" } }" }
  delete: { rule: "{ $ROLE: { eq: \"ADMIN\" } }" }
) {
  id: ID!
  xid: String! @id
  start: Float!
  end: Float!
  text: String! @search(by: [regexp])
  paragraph: Paragraph!
  transcript: Transcript!
  phrase: Phrase
}

What I did

This is my query:

query paginateParagraphsWithPhrases() {
  var(func: type(Transcript)) @filter(eq(Transcript.slug, "hp")) {
    paragraphs as Transcript.paragraphs {
      Paragraph.text {
        wordStart as Word.start
        wordEnd as Word.end
        Phrase.words (orderasc: Word.start) {
          phraseWordStart as Word.start
        }
        phraseStart as min(val(phraseWordStart))
      }
      paragraphStart as min(val(wordStart))
      paragraphEnd as max(val(wordEnd))
    }
  }

  paginateParagraphsWithPhrases(func: uid(paragraphs), orderasc: val(paragraphStart), first: 100, offset: 0) {
    id: uid
    transcript: Paragraph.transcript {
      id: uid
    }
    start: val(paragraphStart)
    end: val(paragraphEnd)
    speaker: Paragraph.speaker {
      name: Speaker.name
    }
    text: Paragraph.text, (orderasc: Word.start) { # want to order this by Word.start, or min(val(phraseWordStart)), i.e. first word in the phrase
      dgraph.type
      id: uid
      phraseStart: val(phraseStart)
      start: Word.start
      end: Word.end
      text: Word.text
      xid: Word.xid
      words: Phrase.words (orderasc: Word.start) {
        xid: Word.xid
        start: Word.start
        text: Word.text
        end: Word.end
      }
    }
  }
}

Dgraph metadata

dgraph version

21.03