Skip to main content

​
Required timestamps in queries

Liberator enforces a time-based predicate on all queries. Every query must include a valid, non-zero date or time range. Queries without one return an error and no data. Previously, a query with no time parameters could return an entire dataset. That behavior was removed to prevent unexpectedly large result sets and unnecessary load.
Review saved notebooks, scripts, and scheduled jobs. Any query that omits back_to / as_of (or an explicit time WHERE in raw sql) will fail until updated.
Applies to Liberator 2.1 and later.

​
Valid timestamps

A valid time scope must:
  • Cover a specific date, date range, or datetime range — not zero, null, or empty.
  • Use back_to and as_of for parameter-based queries (Liberator’s point-in-time window). See Understanding as_of and back_to.
  • Fall within the dataset’s available coverage (valid dates outside coverage return an empty result, not a timestamp error).

​
Correct examples

import liberator

# Single day
df = liberator.get_dataframe(liberator.query(
    name="my_dataset",
    symbols=["AAPL"],
    back_to="2026-01-01",
    as_of="2026-01-01",
))

# Date range
df = liberator.get_dataframe(liberator.query(
    name="my_dataset",
    symbols=["AAPL"],
    back_to="2026-01-01",
    as_of="2026-06-30",
))

# Datetime range
df = liberator.get_dataframe(liberator.query(
    name="my_dataset",
    symbols=["AAPL"],
    back_to="2026-06-01 09:30:00",
    as_of="2026-06-01 16:00:00",
))

​
Incorrect examples

# Missing back_to and as_of — error
liberator.query(name="my_dataset", symbols=["AAPL"])

# Null dates — error
liberator.query(name="my_dataset", symbols=["AAPL"], back_to=None, as_of=None)

​
Raw sql queries

Every SELECT must include an explicit WHERE clause with a time-range predicate on muts or another configured time column. See Query parameters — Raw SQL.

​
What the error looks like

The API or client returns an error indicating a time-based predicate is required. This is expected enforcement, not a platform failure.

​
Update existing queries

  1. Find affected calls — Search for liberator.query() without back_to / as_of, or raw sql without a time WHERE.
  2. Choose the right window — Daily jobs: parameterize to yesterday or today. Historical pulls: set explicit start and end dates.
  3. Add parameters — Use the examples above.
  4. Test — Run manually before re-enabling schedules.

​
FAQ

I used to pull a full dataset for reference. What now?
Set back_to and as_of to the dataset’s available range (ask your administrator or check dataset details in the admin portal). Error but I provided dates?
Confirm values are not null/empty, back_to is not after as_of, and dates use YYYY-MM-DD or YYYY-MM-DD HH:MM:SS. Valid dates but empty results?
The timestamp is accepted; the range may fall outside dataset coverage. AI assistant queries?
Assistants must specify a time range. Timeless requests should prompt for dates; otherwise the query fails like any other.

as_of and back_to

Point-in-time semantics for Liberator datasets

Query parameters

Full parameter reference including raw SQL rules