outcaste-io / outserv Goto Github PK

It'd be great if Outserv had built-in timestamp functionality that worked similarly to Neo4j's:

https://neo4j.com/docs/graphql-manual/current/type-definitions/autogeneration/#type-definitions-autogeneration-timestamp

This was requested multiple times on the Dgraph forums:

https://discuss.dgraph.io/t/query-sever-timestamps-in-graphql/7091
https://discuss.dgraph.io/t/slash-dgraph-backend-security-field-validation-and-timestamps/13317

#61

This product roadmap is based on the initial ideas during launch. As I get more feedback from users, this roadmap is bound to change.

• Importers

  • Smart Contracts
  • NFTs
  • Blockchains (which ones, TBD)
  • Bridge
  • Postgres

• GraphQL Query Cache

So, the Todo example works great now that you have added @hasInverse. I even queried the reverse type (not in the example), and it works as expected.

However, the Quickstart Example for product reviews does not work.

schema

type Product {
    productID: ID!
    name: String @search(by: [term])
    reviews: [Review] @hasInverse(field: about)
}

type Customer {
    username: String! @id @search(by: [hash, regexp])
    reviews: [Review] @hasInverse(field: by)
}

type Review {
    id: ID!
    about: Product!
    by: Customer!
    comment: String @search(by: [fulltext])
    rating: Int @search
}

First Mutation

mutation {
  addProduct(input: [
    { name: "GraphQL on Dgraph"},
    { name: "Dgraph: The GraphQL Database"}
  ]) {
    product {
      productID
      name
    }
  }
  addCustomer(input: [{ username: "Michael"}]) {
    customer {
      username
    }
  }
}

No issues here. The problem is with the second mutation.

Second Mutation

mutation {
  addReview(input: [{
    by: {username: "Michael"}, 
    about: { productID: "0x2"}, 
    comment: "Fantastic, easy to install, worked great.  Best GraphQL server available",
    rating: 10}]) 
  {
    review {
      comment
      rating
      by { username }
      about { name }
    }
  }
}

The by and the about are nested values that should be addable based on the @id field. It is, in reality, an upsert. How you want to handle this may be different than dgraph, but it is a very common use case.

Side Note: Not sure if this is important, but the extensions come back empty on add mutations.

(I imagine you want issues posted here, and features posted under outcaste, so correct me if I'm wrong)

J

We need a solution for renaming a field or a type, without having to rewrite the entire data.

P.S. Have a look at how EdgeDB does this: https://www.edgedb.com/showcase/migrations

We currently use Node.JS to run JS/TS scripts. Research if moving to Deno would make more sense.

#61

#61

Example:

union TextContainer = Word | Phrase

type Paragraph {
  id: ID!
  textContainers: [TextContainer]
  words: [Word!] @hasInverse(field: paragraph)
  phrases: [Phrase] @hasInverse(field: paragraph)
}

type Word {
  id: ID!
  text: String! @search(by: [regexp])
  paragraph: Paragraph!
  phrase: Phrase
}

type Phrase {
  id: ID!
  paragraph: Paragraph!
  words: [Word!]! @hasInverse(field: phrase)
}

If I run deletePhrase() to delete one of the Phrase objects referenced in the Paragraph.textContainers union list, then the API returns the following:

{
  "errors": [
    {
      "message": "Non-nullable field 'xid' (type String!) was not present in result from Dgraph.  GraphQL error propagation triggered.",
      "locations": [
        {
          "line": 12,
          "column": 9
        }
      ],
      "path": [
        "paginateParagraphsWithPhrases",
        0,
        "textContainers",
        3,
        "xid"
      ]
    }
  ],
"data": {
    "paginateParagraphsWithPhrases": [
      {
        "id": "0x42024",
        "transcript": {
          "id": "0x41f6a"
        },
        "start": 0.15,
        "end": 4.19,
        "textContainers": [
          {
            "__typename": "Word",
            "xid": "06dd22c3-3f52-4c4b-aa10-dc177f65e9ee",
            "text": "I'm"
          },
          {
            "__typename": "Word",
            "xid": "0174abd9-a708-4b6f-8f08-e6ae8f36c427",
            "text": "I"
          },
          {
            "__typename": "Phrase",
            "words": [
              {
                "xid": "b42505e7-0379-49c1-95de-64f0ff95212f",
                "text": "only"
              },
              {
                "xid": "e5941e95-8078-487b-a750-15e426dcfa0c",
                "text": "joking."
              }
            ]
          },
          null
        ]
      },

Notice that there is a null in the textContainers list. Desired behaviour is for deleting the Phrase to remove it from the list, or at the very least for the API to just ignore null references.

EDIT: The problem I have now is that I have no idea how to remove that null reference from the list. So the data is effectively corrupted as a result of this simple operation.

This is a problem I'm experiencing in 20.03, I'm assuming it's still an issue in Outserv.

So I am following the guide here: https://dgraph.io/docs/graphql/todo-app-tutorial/todo-schema-design/

I add the data, everything works as expected...

When I query:

query {
  queryUser {
    username
    name
    tasks {
      title
      completed
    }
  }
}

No problems, I get the expected results:

{
  "data": {
    "queryUser": [
      {
        "username": "[email protected]",
        "name": "Alice",
        "tasks": [
          {
            "title": "Avoid crowd",
            "completed": true
          },
          {
            "title": "Wash your hands often",
            "completed": true
          },
          {
            "title": "Avoid touching your face",
            "completed": false
          },
          {
            "title": "Stay safe",
            "completed": false
          }
        ]
      }
    ]
  },
  "extensions": {
    "touched_uids": 16
  }
}

However, when I query the Inverse (from the example), I get this error:

query {
  queryTask {
    id
    title
    completed
    user {
        username
    }
  }
}

{
  "errors": [
    {
      "message": "Non-nullable field 'user' (type User!) was not present in result from Dgraph.  GraphQL error propagation triggered.",
      "locations": [
        {
          "line": 6,
          "column": 5
        }
      ],
      "path": [
        "queryTask",
        0,
        "user"
      ]
    },
    {
      "message": "Non-nullable field 'user' (type User!) was not present in result from Dgraph.  GraphQL error propagation triggered.",
      "locations": [
        {
          "line": 6,
          "column": 5
        }
      ],
      "path": [
        "queryTask",
        1,
        "user"
      ]
    },
    {
      "message": "Non-nullable field 'user' (type User!) was not present in result from Dgraph.  GraphQL error propagation triggered.",
      "locations": [
        {
          "line": 6,
          "column": 5
        }
      ],
      "path": [
        "queryTask",
        2,
        "user"
      ]
    },
    {
      "message": "Non-nullable field 'user' (type User!) was not present in result from Dgraph.  GraphQL error propagation triggered.",
      "locations": [
        {
          "line": 6,
          "column": 5
        }
      ],
      "path": [
        "queryTask",
        3,
        "user"
      ]
    }
  ],
  "data": {
    "queryTask": [
      null,
      null,
      null,
      null
    ]
  },
  "extensions": {
    "touched_uids": 16
  }
}

This may be expected if you did not rewrite the @hasInverse directive. As we discussed, and as the notion says written by Anthony and me, it should be rewritten to use @reverse under-the-hood anyway.

J

#61

The addType function, e.g. addThing, addProduct receives an array of items to add. Always throws me that it's named addType (singular) but takes an array as input. I think it'd make sense to rename it to addTypes, e.g. addProducts, addThings, addUsers

As it is currently:

type Author {
	id: ID!
	name: String! @search(by: [hash])
	dob: DateTime
	posts: [Post]
}

type Post {
	postID: ID!
	title: String! @search(by: [term, fulltext])
	text: String @search(by: [fulltext, term])
	datePublished: DateTime
}

addPost(input: [AddPostInput!]!): AddPostPayload

input AddPostInput {
	title: String!
	text: String
	datePublished: DateTime
}

type AddPostPayload {
	post(filter: PostFilter, order: PostOrder, first: Int, offset: Int): [Post]
	numUids: Int
}

mutation {
  addAuthor(input: [{ name: "A.N. Author", posts: []}]) {
    author {
      id
      name
    }
  }
}

Proposed change:

type Author {
	id: ID!
	name: String! @search(by: [hash])
	dob: DateTime
	posts: [Post]
}

type Post {
	postID: ID!
	title: String! @search(by: [term, fulltext])
	text: String @search(by: [fulltext, term])
	datePublished: DateTime
}

addPosts(input: [AddPostsInput!]!): AddPostsPayload

input AddPostsInput {
	title: String!
	text: String
	datePublished: DateTime
}

type AddPostsPayload {
	post(filter: PostFilter, order: PostOrder, first: Int, offset: Int): [Post]
	numUids: Int
}

mutation {
  addAuthors(input: [{ name: "A.N. Author", posts: []}]) {
    author {
      id
      name
    }
  }
}

We directly parse GraphQL queries to the underlying query struct, but then convert it to DQL, before passing it to edgraph. Avoid that conversion to DQL step.

There're a lot of subgraphs that are built for "The Graph". It would be great if we could write a script that can import them automatically to Outserv as well.

Outserv should allow caching the responses of a GraphQL query for a given TTL, specified in the schema.

#61

Outserv really needs a tool or workflow for safely managing migrations.

Ideally it'd have:

• Ability to generate editable migration scripts that can be used to modify the schema and migrate data, or rollback the changes (and the data) if there was a problem
• Scripts would be timestamped in the filenames, and Outserv would be able to run them in sequence, meaning that developers can migrate their local databases with a CLI command after a schema change in the repo
• CLI commands for:
  • generating a new timestamped migration script file and automatically adding it to the migrations directory
  • running the latest migrations
  • rolling back the last run migrations

Existing migration tools for reference:

https://www.edgedb.com/docs/cli/edgedb_migration/index

https://github.com/fauna-labs/fauna-schema-migrate

https://guides.rubyonrails.org/active_record_migrations.html