'bigint' type is a string in query results.

Following query returns a Postgres bigint for the assigned_client_count column (results of COUNT(*).

This is a string, rather than a Number, in the returned results.

{ __cid: '__cid40',
  method: 'select',
  options: undefined,
  bindings: [ 1 ],
  sql: 'select "id", "name", "fluid_account_id", "survey_fid", coalesce(assigned_client_count, 0) as assigned_client_count from (select "survey_id", count(*) as "assigned_client_count" from "client_surveys" group by "survey_id") as "t_counts" right join "fluid_surveys" on "t_counts"."survey_id" = "fluid_surveys"."id" where "fluid_account_id" = $1' }

 [ { id: 6,
    name: 'some name',
    fluid_account_id: 1,
    survey_fid: 69942,
    assigned_client_count: '1' } ]

[sidazhang]

@rhys-vdw , I have the same issue. The work around is add the following 2 lines above require('knex')

var pg = require('pg')
pg.types.setTypeParser(20, 'text', parseInt)

[bendrucker]

So this is a node-postgres design decision and not really one we can abstract over either:

https://github.com/brianc/node-postgres/pull/353

node-postgres exposes its type parser so you can always override its default for bigint and run it against parseInt for all queries.

The problem here is that if you actually had big integers JS can’t natively deal with them. The default allows for full precision and you get the ability to opt out if you don’t need it.