Documents without the field 'a' will match the query predicate {a: null}, but in the aggregation pipeline, documents without the field 'a' will not match the predicate {$eq: ['$a', null]}.

For example, suppose you had the following documents in the collection coll:

{_id: 0}
{_id: 1, a: null}

Then a query with the predicate {a: null} will return both documents. However, the pipeline:

db.coll.aggregate([{$project: {matches: {$eq: ['$a', null]}}}])

produces the following documents:

{_id: 0, matches: false}
{_id: 1, matches: true}

This is inconsistent, and sometimes not the desired behavior. The aggregation $eq operator should treat nulls the same way the query $eq operator does. Or, there should be an $isNull operator, that behaves like the query predicate {$eq: ['field', null]} operator.

There is a workaround; the following pipeline would treat null and missing as equivalent:

db.coll.aggregate([{
  $project: {
    matches: {
      $eq: [
        {$ifNull: ['$a', null]},
        null
      ]
    }
  }
}])