mongodb bulkWrites update behaves differently from normal update

I have a document with an order array that contains ids. Sometimes I want to set specific indexes in that array:

db.getCollection('myCollection').update(
  {
    _id: 'someId',
  },
  {
    $set: {
      'order.1': 'foo2',
    },
  },
);

This works correcly and updates the second array element. If the array is empty, the 0th element is null.

Trying the same with bulkWrite leads to really weird behavior:

db.getCollection('myCollection').bulkWrite([
  {
    updateOne: {
      filter: { _id: 'eager' },
      update: [
        {
          $set: {
            'order.1': 'foo3',
          },
        },
      ],
    },
  },
]);

If the array in the db is empty this does nothing. However, if there are elements in the array, this replaces every array element with an object of the form { '1': 'foo3' }.

Does anyone have any idea why this happens and how to make bulkWrite behave? :D

In the single update query, you're using the regular update with Update Operators like $set. Mongo Playground

But in the bulkWrite example, you're using the aggregation pipeline format [] for updateOne, which interprets the field notations for $set differently - it applies that expression to each element of the array.

Remove the [] so that it applies as a regular update. So:

db.getCollection('myCollection').bulkWrite([
  {
    updateOne: {
      filter: { _id: 'eager' },
      update: { $set: { 'order.1': 'foo3' } },
    },
  },
]);

And wrt "I used an array because I wanted to update multiple fields (with a $set each)" - you can use a single set operation with multiple fields specified. Example from the docs:

{ $set: { "a.2": <new value>, "a.10": <new value>, } }

However, if you need to reference the value of a field in the update, then you will need to use the pipeline notation and change the $set expression to something that would work in an aggregation pipeline.