Inconsistent mongo/postgres unique constraint behavior

Bug report

I have a model:

keystone.createList('Organization', {
        phone: {
            type: Text,
            isUnique: true,
        },
        ...
})

• Mongo: it means that I have only one record with a null phone.
• Postgres: it means that I can have multiple records with a null phone.

CI tests fail with:

      [
        {
          message: 'E11000 duplicate key error collection: main.users index: phone_1 dup key: { phone: null }',
          locations: [ { line: 2, column: 3 } ],
          path: [ 'user' ],
          extensions: {
            code: 'INTERNAL_SERVER_ERROR',
            exception: {
              driver: true,
              name: 'MongoError',
              index: 0,
              code: 11000,
              keyPattern: { phone: 1 },
              keyValue: { phone: null },
              errmsg: 'E11000 duplicate key error collection: main.users index: phone_1 dup key: { phone: null }',

How-to solve

Use sparse index for MongoDB: more info here: https://stackoverflow.com/a/24430345

[pahaz]

I think it’s better to drop unique field options and move it to knexOptions and mongooseOptions. If you don’t want to homogenise Mongo/Postgres behavior.

It’s unexpected for me. And I can’t write Keystone modules for Mongo and Postgres. I should choose one!

[lachlanhawthorne]

Thank you @pahaz !! I was able to get this to work using the following: phoneNumber: { type: Text, isUnique: true, isRequired: false, mongooseOptions: { sparse: true } },