Documentation index


SOURCE_DATE_EPOCH is a standardised environment variable that distributions can set centrally and have build tools consume this in order to produce reproducible output. In practice, SOURCE_DATE_EPOCH specifies the last modification of something, usually the source code, measured in the number seconds since the Unix epoch, ie. January 1st 1970, 00:00:00 UTC.

Before added support to a tool for reading this variable, you should scan through Debian’s checklist to see if you can avoid implementing it: a number of tools already will do this for you. However, if you find that it’s ideal for your use-case, please feel free to jump straight to our published specification.

Reading the variable

Python >= 3.x

import os
import time
import datetime

build_date = datetime.datetime.fromtimestamp(
    int(os.environ.get('SOURCE_DATE_EPOCH', time.time())),

… or with fewer imports and rendering to a string:

import os
import time

date_str = time.strftime(
    time.gmtime(int(os.environ.get('SOURCE_DATE_EPOCH', time.time())))

for zip files:

import os
import time
import zipfile
filetime = max(315532800, int(os.environ.get('SOURCE_DATE_EPOCH', time.time())))
with zipfile.ZipFile("", 'w') as ziparchive:
    file_zipinfo = zipfile.ZipInfo("hello.txt", date_time=time.gmtime(filetime))
    ziparchive.writestr(file_zipinfo, "Hello world\n")

Python >= 2.x

If you still require Python 2.x support, you will need to use the non-recommended datetime.utcfromtimestamp method (more info):

import os
import time
import datetime

build_date = datetime.datetime.utcfromtimestamp(
    int(os.environ.get('SOURCE_DATE_EPOCH', time.time()))

Bash / POSIX shell

For GNU systems:

BUILD_DATE="$(date --utc --date="@${SOURCE_DATE_EPOCH:-$(date +%s)}" +%Y-%m-%d)"

If you need to support BSD date as well you should fallback to trying ther -r seconds timestamp variant:

BUILD_DATE=$(date -u -d "@$SOURCE_DATE_EPOCH" "$DATE_FMT" 2>/dev/null || date -u -r "$SOURCE_DATE_EPOCH" "$DATE_FMT" 2>/dev/null || date -u "$DATE_FMT")


use POSIX qw(strftime);
my $date = strftime("%Y-%m-%d", gmtime($ENV{SOURCE_DATE_EPOCH} || time));


DATE_FMT = +%Y-%m-%d
    BUILD_DATE ?= $(shell date -u -d "@$(SOURCE_DATE_EPOCH)" "$(DATE_FMT)" 2>/dev/null || date -u -r "$(SOURCE_DATE_EPOCH)" "$(DATE_FMT)" 2>/dev/null || date -u "$(DATE_FMT)")
    BUILD_DATE ?= $(shell date "$(DATE_FMT)")

The above will work with either GNU or BSD date, and fallback to ignore SOURCE_DATE_EPOCH if both fails.



… will compile with CMake versions 2.8.11 and higher (released May 2013), but it only respects SOURCE_DATE_EPOCH since version 3.8.0 (April 2017). Note that the final argument UTC is required or the timestamp may vary between timezones.

If you would like to support legacy/archival versions of CMake, you can use this less-preferred variant:

    COMMAND "date" "-u" "-d" "@$ENV{SOURCE_DATE_EPOCH}" "+%Y-%m-%d"
else ()
    COMMAND "date" "+%Y-%m-%d"
endif ()

Note that the above will work only with GNU date; see the POSIX shell example on how to support BSD date.


By deliberate design, Meson does not provide access to environment variables in build files which makes accessing SOURCE_DATE_EPOCH troublesome.

date_exe = find_program('date')
cmd = run_command('sh', '-c', 'echo $SOURCE_DATE_EPOCH')
source_date_epoch = cmd.stdout().strip()
if source_date_epoch == ''
	source_date_epoch = run_command(date_exe, '+%s').stdout().strip()

formatted_date = run_command(date_exe, '-u', '-d', '@' + source_date_epoch, '+%Y-%m-%d').stdout().strip()

The above will work only with GNU date. See the POSIX shell example on how to support BSD date variants.


The SOURCE_DATE_EPOCH argument value is automaticallly propagated from the SOURCE_DATE_EPOCH environment value of the client host, since Docker Buildx v0.10.

The SOURCE_DATE_EPOCH argument value can be captured as an environment value for RUN instructions, by putting an ARG instruction between FROM and RUN:

FROM [...]
RUN [...]

Capturing the SOURCE_DATE_EPOCH argument value is optional. Alternatively, the SOURCE_DATE_EPOCH environment value can be declared inside a RUN instruction too.

FROM [...]
ADD src /src
RUN [...]
  SOURCE_DATE_EPOCH="$(find /src -type f -exec stat -c '%Y' {} + | sort -nr | head -n1)"; \
# for logging validation/edification
  date --date "@$SOURCE_DATE_EPOCH" --rfc-2822; \


Regardless to whether the SOURCE_DATE_EPOCH argument value is captured into Dockerfile with ARG SOURCE_DATE_EPOCH, the argument value is also used for: - the created timestamp in the OCI Image Config

  • the created timestamp in the history objects in the OCI Image Config
  • the org.opencontainers.image.created annotation in the OCI Image Index
  • the timestamp of the files exported with the local exporter - the timestamp of the files exported with the tar exporter

To apply the SOURCE_DATE_EPOCH argument value to the timestamps of the files inside the image, specify rewrite-timestamp=true as an image exporter option:

docker buildx create --use --name buildkit
docker buildx build --output type=image,,push=true,rewrite-timestamp=true .

The rewrite-timestamp option is not set to true by default due to the overhead of rewriting image layers.


RUN apt-get does not automatically consume SOURCE_DATE_EPOCH to install packages from the past snapshot. can be used for reconfiguring /etc/apt/sources.list to use<SOURCE_DATE_EPOCH>/.

Further information

See: - -


#include <errno.h>
#include <limits.h>
#include <stdlib.h>

struct tm *build_time;
time_t now;
char *source_date_epoch;
unsigned long long epoch;
char *endptr;

source_date_epoch = getenv("SOURCE_DATE_EPOCH");
if (source_date_epoch) {
	errno = 0;
	epoch = strtoull(source_date_epoch, &endptr, 10);
	if ((errno == ERANGE && (epoch == ULLONG_MAX || epoch == 0))
			|| (errno != 0 && epoch == 0)) {
		fprintf(stderr, "Environment variable $SOURCE_DATE_EPOCH: strtoull: %s\n", strerror(errno));
	if (endptr == source_date_epoch) {
		fprintf(stderr, "Environment variable $SOURCE_DATE_EPOCH: No digits were found: %s\n", endptr);
	if (*endptr != '\0') {
		fprintf(stderr, "Environment variable $SOURCE_DATE_EPOCH: Trailing garbage: %s\n", endptr);
	if (epoch > ULONG_MAX) {
		fprintf(stderr, "Environment variable $SOURCE_DATE_EPOCH: value must be smaller than or equal to %lu but was found to be: %llu \n", ULONG_MAX, epoch);
	now = epoch;
} else {
	now = time(NULL);
build_time = gmtime(&now);

If you want less verbose code and are happy with the assumptions stated below, you can use

#include <stdlib.h>

time_t now;
char *source_date_epoch;
/* This assumes that the SOURCE_DATE_EPOCH environment variable will contain
   a correct, positive integer in the time_t range */
if ((source_date_epoch = getenv("SOURCE_DATE_EPOCH")) == NULL ||
    (now = (time_t)strtoll(source_date_epoch, NULL, 10)) <= 0)


#include <sstream>
#include <iostream>
#include <cstdlib>
#include <ctime>

  time_t now;
  char *source_date_epoch = std::getenv("SOURCE_DATE_EPOCH");
  if (source_date_epoch) {
    std::istringstream iss(source_date_epoch);
    iss >> now;
    if ( || !iss.eof()) {
      std::cerr << "Error: Cannot parse SOURCE_DATE_EPOCH as integer\n";
  } else {
    now = std::time(NULL);


import (


source_date_epoch := os.Getenv("SOURCE_DATE_EPOCH")
var build_date string
if source_date_epoch == "" {
        build_date = time.Now().UTC().Format(http.TimeFormat)
} else {
        sde, err := strconv.ParseInt(source_date_epoch, 10, 64)
        if err != nil {
                panic(fmt.Sprintf("Invalid SOURCE_DATE_EPOCH: %s", err))
        build_date = time.Unix(sde, 0).UTC().Format(http.TimeFormat)


\date('Y', (int)\getenv('SOURCE_DATE_EPOCH') ?: \time())


  (when (getenv "SOURCE_DATE_EPOCH")
        (getenv "SOURCE_DATE_EPOCH"))))))


let build_date =
    float_of_string (Sys.getenv "SOURCE_DATE_EPOCH")
    Not_found -> Unix.time ()

Java / gradle

def buildDate = System.getenv("SOURCE_DATE_EPOCH") == null ?
  new java.util.Date() :
  new java.util.Date(1000 * Long.parseLong(System.getenv("SOURCE_DATE_EPOCH")))

Javascript / node.js

var timestamp = new Date(process.env.SOURCE_DATE_EPOCH ? (process.env.SOURCE_DATE_EPOCH * 1000) : new Date().getTime());

// Alternatively, to ensure a fixed timezone:

var now = new Date();
if (process.env.SOURCE_DATE_EPOCH) {
  now = new Date((process.env.SOURCE_DATE_EPOCH * 1000) + (now.getTimezoneOffset() * 60000));


now = new Date()
if process.env.SOURCE_DATE_EPOCH
  now = new Date((process.env.SOURCE_DATE_EPOCH * 1000) + (now.getTimezoneOffset() * 60000))


  now =
  now =['SOURCE_DATE_EPOCH'].to_i).gmtime

Note that Ruby’s Datetime.strftime is locale-independent by default.


To get milliseconds since Epoch:

sys.env.get("SOURCE_DATE_EPOCH").map(_.toLong * 1000)

To get a java.util.Date:

  .map(sde => new java.util.Date(sde.toLong * 1000))


Using the chrono crate:

use chrono::{TimeZone, Utc};
use std::env;

let now = match env::var("SOURCE_DATE_EPOCH") {
    Ok(val) => { Utc.timestamp_opt(val.parse::<i64>().unwrap(), 0).unwrap() }
    Err(_) => Utc::now(),


use chrono::{DateTime, NaiveDateTime, Utc};
use std::env;

let now = match env::var("SOURCE_DATE_EPOCH") {
    Ok(val) => {
        let naive = NaiveDateTime::from_timestamp(val.parse::<i64>().unwrap(), 0);
        let datetime: DateTime<Utc> = DateTime::from_utc(naive, Utc);
    Err(_) => Utc::now(),


Add the following property in the pom.xml file:



Set the following properties for the Zip Task that creates the .jar file:

// Normalizes the ZIP and JAR archives
tasks.withType(Zip) {
   preserveFileTimestamps = false
   reproducibleFileOrder = true


Builds of OpenJDK version 19 or later support the following new option in the jar and jmod commands:

   The timestamp in ISO-8601 extended offset date-time with optional
   time-zone format, to use for the timestamp of the entries,
   e.g. "2022-02-12T12:30:00-05:00".


import java.time.Instant
import java.time.temporal.ChronoUnit

def buildInstant =
def sourceDateEpoch = System.getenv("SOURCE_DATE_EPOCH")
if (sourceDateEpoch != null) {
   def epochSeconds = Long.parseLong(sourceDateEpoch)
   buildInstant = Instant.ofEpochSecond(epochSeconds)
// Creates the timestamp in UTC using the ISO 8601 extended format.
def extendedTimestamp = buildInstant.toString()

Last-resort using faketime

'’As a last resort to be avoided where possible’’ (e.g. if the upstream tool is too hard to patch, or too time-consuming for you right now to patch, or if they are being uncooperative or unresponsive), package maintainers may try something like the following:


# Depends: faketime
# Eventually the upstream tool should support SOURCE_DATE_EPOCH internally.
test -n "$SOURCE_DATE_EPOCH" || { echo >&2 "$0: SOURCE_DATE_EPOCH not set"; exit 255; }
exec env NO_FAKE_STAT=1 faketime -f "$(TZ=UTC date -d "@$SOURCE_DATE_EPOCH" +'%Y-%m-%d %H:%M:%S')" /usr/bin/a2x "$@"


export PATH := $(CURDIR)/debian/strip-nondeterminism:$(PATH)


Build-Depends: faketime

But please be aware that this does not work out of the box with pbuilder on Debian 7 Wheezy, see #778462 against faketime and #700591 against pbuilder (fixed in Jessie, but not Wheezy). Adding an according hook to /etc/pbuilder/hook.d which mounts /run/shm inside the chroot should suffice as local workaround, though.

TODO: document some other nicer options. Generally, all invocations of date(1) need to have a fixed TZ environment set.

NOTE: faketime BREAKS builds on some archs, for example hurd. See #778462 for details.

Setting the variable


In Debian, this is automatically set to the same time as the latest entry in debian/changelog, i.e. the same as the output of dpkg-parsechangelog -SDate.

  1. For packages using debhelper versions >= 9.20151004, this variable is automatically exported during builds, so you probably don’t need to change anything. One exception is if your debian/rules needs this variable in non-debhelper parts, in which case you can try (3) or (4).

  2. For packages using CDBS, versions >= 0.4.131 this variable is also exported automatically during builds, so no changes are needed.

  3. With dpkg >= 1.18.8 you can either include /usr/share/dpkg/ or include /usr/share/dpkg/ This was added in Debian bug #824572.

  4. If none of the above options are good (you should have a ‘‘very good reason’’) then package maintainers may set and export this variable manually in debian/rules:

export SOURCE_DATE_EPOCH ?= $(shell dpkg-parsechangelog -STimestamp)

If you need/want to support dpkg versions earlier than 1.18.8:

export SOURCE_DATE_EPOCH ?= $(shell dpkg-parsechangelog -SDate | date -f- +%s)

If you need/want to support dpkg versions earlier than 1.17.0:

export SOURCE_DATE_EPOCH ?= $(shell dpkg-parsechangelog | grep -Po '^Date: \K.*' | date -f- +%s)

This snippet is believed to work on dpkg versions as far back as 2003.


To set SOURCE_DATE_EPOCH to the last modification of a Git repository you can use git log. For example:

export SOURCE_DATE_EPOCH=$(git log -1 --pretty=%ct)

Tool support

We are persuading upstream tools to support SOURCE_DATE_EPOCH directly. You may help by writing patches for these tools; please add links to the bug reports here so we know, and to act as an example resource for future patch writers.


Or you can search in all Debian sources.

More detailed discussion

Sometimes developers of build tools do not want to support SOURCE_DATE_EPOCH, or they will tweak the suggestion to something related but different. We really do think the best approach is to use SOURCE_DATE_EPOCH exactly as-is described above in our proposal, without any variation. Here we explain our reasoning versus the arguments we have encountered.

(See Standard Environment Variables for general arguments.)

“Lying about the time” / “violates language spec”

This argument arises when the tool processes some input which contains a static instruction to the effect of “get_current_time()”. The input has a specification that defines what this means. The tool executes this instruction, then embeds the result in the output. It is argued that SOURCE_DATE_EPOCH would break these semantics and violate the specification.

In most cases, this argument places too much weight on the absoluteness of time. Regardless of what any specification says, the user can set their own system clock and achieve an effect similar to SOURCE_DATE_EPOCH. Note: Setting the system clock is not enough for ‘‘reliable’’ reproducible builds

  • we need SOURCE_DATE_EPOCH for long-running build processes that take varying amounts of time. If the tool was run near the end of the process, then merely setting the system clock would not make timestamps here reproducible. It is not up to the tool to judge whether the user is lying with their system clock, and likewise, it is not up to the tool to judge whether SOURCE_DATE_EPOCH is a “lie” or not.

For all intents and purposes, if the user has set SOURCE_DATE_EPOCH then they are taking a position that “this is the current time; please use this instead of whatever clock you normally use”. Yes, the project developer wrote “get_current_time()” but I as the user, by setting SOURCE_DATE_EPOCH, am choosing to override this with my own idea of what time it is. Please execute the build as if the current time was SOURCE_DATE_EPOCH. FOSS software should generally prefer to respect end-users’ wishes rather than developers’ wishes. (And in practise, we haven’t seen ‘‘any’’ instance where a project developer really really prefers “time of build” over “modtime of source”.)

In conclusion, the tool may choose to ignore SOURCE_DATE_EPOCH for other reasons, but to judge that this is a ‘‘lie’’ is to disrespect the user’s wishes. Furthermore, choosing to support this is unlikely to ‘‘actually’’ violate any specifications, since they generally don’t define “current”. This does not take into account, if the specification needs to interoperate consistently with other programs in a strong cryptographic ledger protocol where time values ‘‘must’’ be consistent across multiple entities. However this scenario is unlikely to apply, in the context of build tools where SOURCE_DATE_EPOCH would be set.)

Many tools allow the user to override the “current” date - e.g. -D__TIME__=xxx, \\year=yyy, etc. In these cases, it makes even less sense to ignore SOURCE_DATE_EPOCH for data integrity reasons - you ‘‘already’’ have a mechanism where the user can “lie” or “break the spec”; SOURCE_DATE_EPOCH would just be adding an extra mechanism that makes it easier to do this globally across all tools.

If for some reason you’re still conflicted on suddenly changing the meaning of your “now()” function and desire another switch other than SOURCE_DATE_EPOCH being set or not, the texlive project came up with the variable FORCE_SOURCE_DATE; when that environment variable is set to 1 cases that wouldn’t normally obey SOURCE_DATE_EPOCH will do. We strongly discourage the usage of such variable; SOURCE_DATE_EPOCH is meant to be already a flag forcing a particular timestamp to be used.

OTOH, one alternative we can agree with, that also avoids SOURCE_DATE_EPOCH, would be to translate the static instruction “get_current_time()” from the input format to ‘‘an equivalent instruction’’ in the output format, if the output format supports that.

Interaction of SOURCE_DATE_EPOCH with automatic rebuilds

When build dependencies change, it is sometimes necessary to rebuild a package without otherwise changing the source. This is especially the case when a security bug is embedded from the build dependency into the build output, as a rebuild is required to fix the security issue. There are, however, many are other situations. Whilst rebuilds of this kind are a manual process in many distributions, some do this automatically.

Debian only performs these rebuilds via a manual developer request called Binary Non-Maintainer Upload (binNMU). For this, SOURCE_DATE_EPOCH is incremented by 1 to avoid creating a situation where two build results contain files with the same name and the same modification timestamp but a different content. Not increasing the date breaks rsync without the --checksum option or indeed any transfer/sync/backup process that relies on the mtime of files to change to detect underlying content changes. While it is not generally safe to rely on modification times (and thus also not to backup systems without --checksum), we should not unnecessarily break things.

OpenSUSE performs such rebuilds automatically and also uses something called “build tree pruning”. Build tree pruning is to discard the output when it is the same and thus not trigger such rebuilds on packages that build depend on the discarded output. Normally this works, but if a build e.g. embeds SOURCE_DATE_EPOCH in its output, then the output changes every time such a rebuild happens, which can be very often. This is to be avoided as updating packages without necessity is too expensive. (The correct solution would, of course, be to never embed SOURCE_DATE_EPOCH as so far, no use-case was found were something else is not better, like instead embed a major version number. However convincing everyone and changing every instance of this is an impractical task.)

The following achieves both detecting rebuilds that did not change, despite a changed build dependency and having an incremented mtime on files in the output:

  • Use the old SOURCE_DATE_EPOCH that was not incremented for the inner build script.
  • Use the incremented SOURCE_DATE_EPOCH for the outer build script which uses it to set the file date after the inner build script is done.
  • Instead of clamping the file date to SOURCE_DATE_EPOCH, set it exactly. This avoids bugs where one of the inner build scripts sets the file date to something earlier, causing two builds to output a file with same date and name, but different content.
  • Ignore the file date when comparing build output to decide whether to discard for build tree pruning.

History and alternative proposals

1 and the surrounding messages describe the initial motivation behind this, including an evaluation of how different programming languages handle date formats.

We do not have a proposal that includes anything resembling a “time zone”. Developing such a standard requires consideration of various issues:

Intuitive and naive ways of handling human-readable dates, such as the POSIX date functions, are highly flawed and freely mix implicit not-well-defined calendars with absolute time. For example, they don’t specify they mean the Gregorian calendar, and/or don’t specify what to do with dates before when the Gregorian calendar was introduced, or use named time zones that require an up-to-date timezone database (e.g. with historical DST definitions) to parse properly.

Since this is meant to be a universal standard that all tools and distributions can support, we need to keep things simple and precise, so that different groups of people cannot accidentally interpret it in different ways. So it is probably unwise to try to standardise anything that resembles a named time zone, since that is very very complex.

One likely candidate would be something similar to the git internal timestamp format, see man git-commit:

It is

We already have SOURCE_DATE_EPOCH so the time zone offset could be placed in SOURCE_DATE_TZOFFSET or something like that. But all of this needs further discussion.

Other non-standard variables that we haven’t yet agreed upon, use at your own risk:

export SOURCE_DATE_TZOFFSET = $(shell dpkg-parsechangelog -SDate | tail -c6)
export SOURCE_DATE_RFC2822 = $(shell dpkg-parsechangelog -SDate)
export SOURCE_DATE_ISO8601 = $(shell python -c 'import time,email.utils,sys;t=email.utils.parsedate_tz(sys.argv[1]);\
print(time.strftime("%Y-%m-%dT%H:%M:%S",t[:-1])+"{0:+03d}{1:02d}".format(t[-1]/3600,t[-1]/60%60));' "$(SOURCE_DATE_RFC2822)")

The ISO8601 code snippet is complex, in order to preserve the same timezone offset as in the RFC2822 form. If one is OK with stripping out this offset, i.e. forcing to UTC, then one can just use date -u instead. However, this then contains the same information as the unix timestamp, but the latter is generally easier to work with in nearly all programming languages.

Documentation index