When trying to use ConnectionFromMongoCursor I'm getting the error:

TypeError: Cannot read property 'model' of undefined ... graphql-mongoose-loader/lib/ConnectionFromMongoCursor.js:53:34

Or this line:

clonedCursor = cursor.model.find().merge(cursor);

I've tried debugging and believe it's to do with the getTotalCount call but since this is async wrapped in the regeneratorRuntime its proving difficult. Is this a problem for anyone else?

Thank you.

results.forEach

Any plans to upgrade to Dataloader v2.0.0?

1. Imagine a case where initially we fetched 20 users (sorted according to createdAt: -1). It got us the ids: [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20]

2. rerun the app and fetch only 10 users initially. It will give us ids as expected: [1,2,3,4,5,6,7,8,9,10]

3. Delete the user with id: 2 and Load more 10 users

4. It will skip the user with id: 11 and fetch the users with ids: [12,13,14,15,16,17,18,19,20,21]

I have not tested the case with addition; i believe it may have similar effect in the opposite direction. Not sure.

I'm upgrading from 2 to 4 which assumes breaking changes yet I have to dig through commits to find out if this is true yet what they are. Apologies if there is actually a log somewhere but I cannot find one in releases nor the readme or changelog.md

Currently we are projecting only the id field and passing that to the loader:
https://github.com/entria/graphql-mongoose-loader/blob/master/src/ConnectionFromMongoAggregate.js#L54

We should allow the user to choose to receive the raw values, by adding a raw flag.

• add new param estimated that will use estimatedDocumentCount instead of countDocuments to make total count processing faster
• add on both mongo and aggregation cursors

https://mongoosejs.com/docs/api.html#model_Model.estimatedDocumentCount

Currently package.json says :

  "peerDependencies": {
    "graphql": "^0.10.0 || ^0.11.0 || ^0.12.0 || ^0.13.0"
  },

but I don't see any code directly relying on the graphql package...
On the contrary, graphql-relay and mongoose are require-d but not mentioned as peerDependencies.

use babel-present-env

use node 8

so we don't need to transpile async await

I have got Post Schema which contains Comment Ids through References

const PostSchema = new Schema({
  title: { type: String },
  details: { type: String },
  comments: [{
    type: Schema.Types.ObjectId,
    ref: DB_MODELS.COMMENT,
  }],
});

PostSchema.statics.getComments = function(postId) {
  return this.findById(postId)
    .populate('comments')
    .then(post => post.comments);
};

const Post = mongoose.model('post', PostSchema);
export default Post;

My GraphQL type for the same:

const GraphQLPost = new GraphQLObjectType({
  name: 'Post',
  fields: {
    id: globalIdField('Post'),
    _id: { type: GraphQLID },
    title: { type: GraphQLString },
    details: { type: GraphQLString },
    comments: {
      type: CommentConnection,
      args: {
        ...connectionArgs,
      },
      resolve: async (post, args) => {
        console.log('resolving comments inside GraphQLPost..', args);
        const comments = await Post.getComments(post._id);
        if (!comments) {
          return [];
        }
        return connectionFromArray(comments, args);
      },
    },
  },
  interfaces: () => [nodeInterface],
});

export default GraphQLPost;

How i can use this module to load subdocuments for a document i.e. comments for a Post?

There are obvious problems with my existing approach as it will fetch all the comments for a post. That's why i want to use this module to paginate + use DataLoader.

https://launchpad.graphql.com/

Jest test not running. Does not show any errors. My OS is Ubuntu 19, I use node v12.6.0, yarn v1.17.3 and jest 23.5.0.

Below is the print of my terminal:

after some upgrades we know have a lot of type errors to fix

https://github.com/entria/graphql-mongoose-loader/blob/master/src/ConnectionFromMongoCursor.js#L7

https://www.npmjs.com/package/@entria/graphql-mongoose-loader

Hey guys,

Currently the mongooseLoader works only with the _id field, however it would be really useful to load data based on other keys as well (e.g. loading users by username).

It would only require a key field parameter in the mongooseLoader function. A possible implementation could be the folowing:

export default async function mongooseLoader(
  model: Mongoose$Document,
  keys: ReadonlyArray<string>,
  lean: boolean = true,
  keyField: string = '_id'
) {
  const results = lean ? await model.find({ [keyField]: { $in: keys } }).lean() : await model.find({ [keyField]: { $in: keys } });

  return normalizeResults(keys, keyField, cacheKeyFn)(results);
}

The lean and keyField parameters could be moved to an options object, too.

Please share your thoughts on it!

I am using mongodb native driver. Would you mind creating graphql-mongodb-loader as well? Thanks!

What do you suggest should be the response in that case?

I think it should be an empty set of edges.

First of all, thanks for this package, you save me from a long boilerplate code.
BTW, I encounter a little problem using this package:

if you try to use this module without babel-polyfill package you encounter this error:
error: ReferenceError: regeneratorRuntime is not defined

For a workaround you need to npm install babel-polyfill and require('babel-polyfill') in your app entry point.

Bye

If you could give some introduction to what actually it is solving would be great.

enable using .lean() to improve performance

http://www.tothenew.com/blog/high-performance-find-query-using-lean-in-mongoose-2/