date_sub

Applies to VAST Cluster 5.5.

Functions › Date and time functions › Differences. Returns the number of complete parts, such as whole months or whole days, between two dates, timestamps or times: from 31 January to 1 February is zero months. datesub is the same function. Despite its name, it doesn't subtract an interval from a date: for that, use -.

Syntax

date_sub(part, start, end)
datesub(part, start, end)

Returns: BIGINT: positive when end is after start, negative when it's before.

Aliases: datesub.

Set up the examples

The examples connect with the ADBC driver, as in VAST Query Engine (ADBC), and work in a schema called docs_sql_fn_date_time_date_sub. Set QE_ENDPOINT, AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY and VASTDB_BUCKET (SQL reference), and run the blocks in order in one Python session. Below each block is the output you should see.

This block connects, defines the run() and plan() helpers, and creates a table of five events, each with a date (d), a timestamp (ts) and a time of day (t). Row 4 has no values:

import os
import re

import adbc_driver_manager
import adbc_driver_manager.dbapi as dbapi
import adbc_driver_vastdb

BUCKET = os.environ["VASTDB_BUCKET"]
SCHEMA = f'"{BUCKET}/docs_sql_fn_date_time_date_sub"'

conn = dbapi.connect(
    driver=adbc_driver_vastdb.get_driver_path(),
    db_kwargs={
        "vast.db.endpoint": os.environ["QE_ENDPOINT"],
        "vast.db.access_key": os.environ["AWS_ACCESS_KEY_ID"],
        "vast.db.secret_key": os.environ["AWS_SECRET_ACCESS_KEY"],
    },
    autocommit=True,
)

def server_message(e):
    """The server's message in a driver error, without the request ID or the echoed statement."""
    text = re.sub(r"\[req=0x[0-9a-f]+\]: ", "", str(e))
    found = (re.search(r'message: "(.*?)", request_id', text, re.S)
             or re.search(r"description: (.*)\)", text, re.S))
    message = found.group(1) if found else text.strip().splitlines()[-1]
    message = re.sub(r"^Plan query: '.*?' failed with: ", "", message, flags=re.S)
    message = re.split(r"\\n|\n", message)[0].replace('\\"', '"')
    return message.replace(BUCKET, "<bucket>")

def run(sql):
    """Run one statement. Print the rows it returns, or the error."""
    try:
        with conn.cursor() as cur:
            cur.execute(sql)
            rows = cur.fetchall() if cur.description else []
    except (adbc_driver_manager.Error, OSError) as e:
        print(f"{type(e).__name__}: {server_message(e)}")
    else:
        print(rows)

def plan(sql):
    """Print the plan without its ~N Rows lines: the table's row count, which can still be 0 just after an insert."""
    with conn.cursor() as cur:
        cur.execute("EXPLAIN " + sql)
        steps = cur.fetchall()[0][1]
    print("\n".join(line for line in steps.splitlines() if "Rows" not in line))

TABLE = f"{SCHEMA}.events"
run(f"CREATE SCHEMA IF NOT EXISTS {SCHEMA}")
run(f"CREATE TABLE IF NOT EXISTS {TABLE} (id BIGINT, d DATE, ts TIMESTAMP, t TIME)")
run(f"INSERT INTO {TABLE} VALUES "
    "(1, DATE '2026-02-28', TIMESTAMP '2026-02-28 23:15:00', TIME '23:15:00'), "
    "(2, DATE '2026-03-01', TIMESTAMP '2026-03-01 08:30:45.123456', TIME '08:30:45.123456'), "
    "(3, DATE '2024-12-30', TIMESTAMP '2024-12-30 12:00:00', TIME '12:00:00'), "
    "(4, NULL, NULL, NULL), "
    "(5, DATE '2024-02-29', TIMESTAMP '2024-02-29 00:00:00', TIME '00:00:00')")

[]
[]
[(5,)]

What it does

date_sub counts how many whole parts fit between start and end, and drops the remainder. From 2026-01-31 to 2026-02-01 is zero months and one day; from 2025-12-31 to 2026-01-01 is zero years. For timestamps the time counts too: from TIMESTAMP '2026-01-31 12:00:00' to TIMESTAMP '2026-02-28 11:59:59' is 27 days, one second short of 28.

A month counts as complete when adding it lands on or before end, and adding a month to the end of a month gives the end of the shorter month. So from 31 January to 28 February is one complete month, and from 29 February 2024 to 28 February 2025 is one complete year.

date_diff counts boundaries instead. The two differ whenever a period is incomplete:

On TIME values, date_sub counts within one day, with the time parts only: from TIME '08:00:00' to TIME '09:30:00' is one hour.

NULLs and edge cases

The same cases, as the cluster returns them. They use the table and helpers from Set up the examples:

run(f"SELECT id, date_sub('day', d, DATE '2026-12-31') FROM {TABLE} WHERE id = 4")
run("SELECT date_sub(NULL, DATE '2026-01-01', DATE '2026-02-01'), date_sub('day', DATE '2026-01-01', DATE 'infinity'), "
    "date_sub('day', DATE '2026-01-01', '2026-02-01'), "
    "date_sub('second', TIMESTAMP_NS '2026-03-01 00:00:00.999999999', TIMESTAMP_NS '2026-03-01 00:00:01')")
run("SELECT date_sub('yyyy', DATE '2024-02-29', DATE '2026-03-01')")
run("SELECT date_sub('nanosecond', DATE '2024-02-29', DATE '2026-03-01')")
run("SELECT date_sub('era', DATE '2024-02-29', DATE '2026-03-01')")
run("SELECT date_sub('day', TIME '08:00:00', TIME '09:00:00')")
run("SELECT date_sub('hour', TIME '08:00:00', TIMESTAMP '2026-03-01 09:00:00')")
run("SELECT date_sub('day', '2026-01-01', '2026-02-01')")
run("SELECT date_sub('day', 123, DATE '2026-12-31')")

[(4, None)]
[(None, None, 31, 0)]
OSError: extract specifier "yyyy" not recognized
OSError: extract specifier "nanosecond" not recognized
OSError: Specifier type not implemented for DATESUB
OSError: "time" units "day" not recognized
InternalError: Binder Error: No function matches the given name and argument types 'date_sub(STRING_LITERAL, TIME, TIMESTAMP)'. You might need to add explicit type casts.
InternalError: Binder Error: Could not choose a best candidate function for the function call "date_sub(STRING_LITERAL, STRING_LITERAL, STRING_LITERAL)". In order to select one, please add explicit type casts.
InternalError: Binder Error: No function matches the given name and argument types 'date_sub(STRING_LITERAL, INTEGER_LITERAL, DATE)'. You might need to add explicit type casts.

The comparison with date_diff from What it does, the month-end and leap-day cases, and a TIME difference:

run("SELECT date_sub('month', DATE '2026-01-31', DATE '2026-02-01'), date_diff('month', DATE '2026-01-31', DATE '2026-02-01'), "
    "date_sub('month', DATE '2026-01-31', DATE '2026-03-01'), date_diff('month', DATE '2026-01-31', DATE '2026-03-01'), "
    "date_sub('year', DATE '2025-12-31', DATE '2026-01-01'), date_diff('year', DATE '2025-12-31', DATE '2026-01-01')")
run("SELECT date_sub('hour', TIMESTAMP '2026-03-01 08:59:00', TIMESTAMP '2026-03-01 09:01:00'), "
    "date_diff('hour', TIMESTAMP '2026-03-01 08:59:00', TIMESTAMP '2026-03-01 09:01:00'), "
    "date_sub('year', DATE '2024-02-29', DATE '2025-02-28'), date_diff('year', DATE '2024-02-29', DATE '2025-02-28')")
run("SELECT date_sub('month', DATE '2026-01-31', DATE '2026-02-28'), "
    "date_sub('day', TIMESTAMP '2026-01-31 12:00:00', TIMESTAMP '2026-02-28 11:59:59'), "
    "datesub('hour', TIME '08:00:00', TIME '09:30:00')")

[(0, 1, 1, 2, 0, 1)]
[(0, 1, 1, 1)]
[(1, 27, 1)]

Examples

The whole days from each event to the end of 2026, the whole hours from each timestamp to 1 March 2026, and the whole months to the end of 2026:

run(f"SELECT id, date_sub('day', d, DATE '2026-12-31'), datesub('hour', ts, TIMESTAMP '2026-03-01 00:00:00'), "
    f"date_sub('month', ts, TIMESTAMP '2026-12-31 00:00:00') FROM {TABLE} ORDER BY id LIMIT 10")

[(1, 306, 0, 10), (2, 305, -8, 9), (3, 731, 10212, 24), (4, None, None, None), (5, 1036, 17544, 34)]

The whole minutes from each time of day to noon:

run(f"SELECT id, date_sub('minute', t, TIME '12:00:00') FROM {TABLE} ORDER BY id LIMIT 10")

[(1, -675), (2, 209), (3, 0), (4, None), (5, 720)]

Keep only the events less than 10 whole months before the end of 2026:

run(f"SELECT id FROM {TABLE} WHERE date_sub('month', d, DATE '2026-12-31') < 10 ORDER BY id LIMIT 10")

[(2,)]

How it runs in VAST

The Query Engine runs on the cluster's CNodes, and a query on a table starts with a scan in VAST DataBase (VAST_SCAN). VAST DataBase can evaluate a comparison of a column with a constant, but not a function of a column, so date_sub() is always computed by the Query Engine:

• In the SELECT list, the scan reads the columns and the Query Engine computes date_sub() for each row it returns.
• In WHERE, the scan sends the columns for every row, and a FILTER step in the Query Engine computes date_sub() and keeps the rows that match. No rows are skipped in storage.

The plan of the WHERE example shows the FILTER step above the scan:

plan(f"SELECT id FROM {TABLE} WHERE date_sub('month', d, DATE '2026-12-31') < 10")

┌───────────────────────────┐
│         PROJECTION        │
│    ────────────────────   │
│             #1            │
│                           │
└─────────────┬─────────────┘
┌─────────────┴─────────────┐
│           FILTER          │
│    ────────────────────   │
│   (date_sub('month', d,   │
│ '2026-12-31'::DATE) < 10) │
│                           │
└─────────────┬─────────────┘
┌─────────────┴─────────────┐
│         VAST_SCAN         │
│    ────────────────────   │
│    Function: VAST_SCAN    │
│                           │
│        Projections:       │
│             d             │
│             id            │
│                           │
└───────────────────────────┘

Make it faster

• Compare the column with a computed constant. "Less than one whole year before the end of 2026" is ts > TIMESTAMP '2026-12-31 00:00:00' - INTERVAL 1 YEAR. That condition is pushed down to VAST DataBase, so rows that don't match never leave storage; date_sub('year', ts, ...) < 1 isn't. Both return the same rows, and the plan of the first has no FILTER step:

run(f"SELECT id FROM {TABLE} WHERE date_sub('year', ts, TIMESTAMP '2026-12-31 00:00:00') < 1 ORDER BY id LIMIT 10")
run(f"SELECT id FROM {TABLE} WHERE ts > TIMESTAMP '2026-12-31 00:00:00' - INTERVAL 1 YEAR ORDER BY id LIMIT 10")
plan(f"SELECT id FROM {TABLE} WHERE ts > TIMESTAMP '2026-12-31 00:00:00' - INTERVAL 1 YEAR")

[(1,), (2,)]
[(1,), (2,)]
┌───────────────────────────┐
│         VAST_SCAN         │
│    ────────────────────   │
│    Function: VAST_SCAN    │
│      Projections: id      │
│                           │
└───────────────────────────┘

• For whole days between dates, subtract the dates. DATE '2026-12-31' - d gives the same number as date_sub('day', d, DATE '2026-12-31'); to filter on it, write d > DATE '2026-12-31' - 300 (-).

Limitations and errors

• It doesn't subtract an interval. A number in place of the start date, as in date_sub('day', 1, DATE '2026-03-01'), fails with No function matches the given name and argument types. To go back in time, use -: DATE '2026-03-01' - 1 or ts - INTERVAL 1 MONTH.
• No time-zone support. On TIMESTAMP WITH TIME ZONE values, such as now(), date_sub fails with an Extension Autoloading Error. Cast the values to TIMESTAMP first; see time zones.
• No nanoseconds. Use 'microseconds'.

run("SELECT date_sub('day', 1, DATE '2026-03-01')")
run("SELECT DATE '2026-03-01' - 1, TIMESTAMP '2026-03-31 08:00:00' - INTERVAL 1 MONTH")

InternalError: Binder Error: No function matches the given name and argument types 'date_sub(STRING_LITERAL, INTEGER_LITERAL, DATE)'. You might need to add explicit type casts.
[(datetime.date(2026, 2, 28), datetime.datetime(2026, 2, 28, 8, 0))]

Similar functions

run("SELECT date_sub('month', DATE '2026-01-31', DATE '2026-03-01'), date_diff('month', DATE '2026-01-31', DATE '2026-03-01'), "
    "age(DATE '2026-03-01', DATE '2026-01-31'), DATE '2026-03-01' - DATE '2026-01-31'")

[(1, 2, MonthDayNano(months=1, days=1, nanoseconds=0), 29)]

Clean up:

run(f"DROP TABLE IF EXISTS {TABLE}")
run(f"DROP SCHEMA IF EXISTS {SCHEMA}")
conn.close()

[]
[]

See also

• Date and time functions: all the date and time functions.
• Where filters run and pushdowns.
• EXPLAIN: how to read a plan.