Is there a way to set the batch size in batch processing?

[sesmi123]

Hi,

I am currently trying out the batch processing for N+1 query problem. It is working as expected but I would like to know how the batch size is controlled. Can the batch size be explicitly set?

Any further explanation on how the batching internally works is appreciated.

Thankyou

[PatrykWalach]

I looked at the implementation and it's not possible, although could be after minor change.
For now I've came up with a workaround:

import { batchDelegateToSchema } from "@graphql-tools/batch-delegate";
import { OperationTypeNode } from "graphql";
import { MeshContext, Resolvers } from "./.mesh";

const resolvers: Resolvers = {
  Book: {
    media: {
      selectionSet: `{ mediaId }`,
      async resolve(root, args, context, info) {
        const schema = {
          schema: context.Service.rawSource.schema,
          executor: context.Service.rawSource.executor,
        };

        return batchDelegateToSchema({
          schema,
          operation: OperationTypeNode.QUERY,
          fieldName: "getMediaByIds",
          key: root.mediaId,
          dataLoaderOptions: {
            maxBatchSize: 50,
          },
          context,
          info,
        });
      },
    },
  },
};

export default resolvers;

[vik-osteon]

The original question is still very much relevant. Is the WA like the above a recommended way to proceed?

PS: @sesmi123 thanks heaps for the snippet above, I've spent many hours trying to solve similar problem and just couldnt get batchDelegateToSchema to work in a Mesh setup.

I'll add to the question: if to use configuration-based batch resolver (like in https://the-guild.dev/graphql/mesh/docs/guides/batching), how should null values be accounted for in keyField ? Right now ("@graphql-mesh/cli": "^0.89.0" and "@graphql-mesh/graphql": "^0.97.0") it seems to be failing on the first null it encounters.

Any information will be very much appreciated. Thanks!