Time Zone Support

Introduction

Starting with C++20 the <chrono> library has support for time zones. These are available in the IANA Time Zone Database. This page describes the design decisions and trade-offs made to implement this feature. This page contains several links with more information regarding the contents of the IANA database, this page assumes the reader is familiar with this information.

Which version of the Time Zone Database to use

The data of the database is available on several platforms in different forms:

  • Typically Unix systems ship the database as TZif files. This format has 3 versions and the time_zone_link information is not always available. If available, they are symlinks in the file system. These files don’t provide the database version information. This information is needed for the functions std::chrono:: remote_version() and std::chrono::reload_tzdb().

  • On several Unix systems the time zone source files are available. These files are stored in several regions, mainly the continents. This file contains a large amount of comment with historical information regarding time zones. The format is documented in the IANA documentation and in the man page of zic. The disadvantage of this version is that at least Linux versions don’t have the database version information. This information is needed for the functions std::chrono:: remote_version() and std::chrono::reload_tzdb().

  • On Linux systems tzdata.zi is available. This contains the same information as the source files but in one file without the comments. This file uses the same format as the sources, but shortens the names. For example Rule is abbreviated to R. This file contains the database version information.

The disadvantage of the TZif format (which is a binary format) is that it’s not possible to get the proper time_zone_link information on all platforms. The time zone database version number is also missing from TZif files. Since the time zone database is supposed to contain both these informations, TZif files can’t be used to create a conforming implementation.

Since it’s easier to parse one file than a set of files we decided to use the tzdata.zi. The other benefit is that the tzdata.zi file contains the database version information needed for a conforming implementation.

The tzdata.zi file is not available on all platforms as of August 2023, so some vendors will need to make changes to their platform. Most vendors already ship the database, so they only need to adjust the packaging of their time zone package to include the files we require. One notable exception is Windows, where no IANA time zone database is provided at all. However it’s possible for Windows packagers to add these files to their libc++ packages. The IANA databases can be downloaded.

An alternative would be to ship the database with libc++, either as a file or compiled in the dylib. The text file is about 112 KB. For now libc++ will not ship this file. If it’s hard to get vendors to ship these files we can reconsider based on that information.

Leap seconds

For the leap seconds libc++ will use the source file leap-seconds.list. This file is easier to parse than the leapseconds file. Both files are present on Linux, but not always on other platforms. Since these platforms need to change their packaging for tzdata.zi, adding two instead of one files seems a small change.

Updating the Time Zone Database

Per [time.zone.db.remote]/1

The local time zone database is that supplied by the implementation when the
program first accesses the database, for example via current_zone(). While the
program is running, the implementation may choose to update the time zone
database. This update shall not impact the program in any way unless the
program calls the functions in this subclause. This potentially updated time
zone database is referred to as the remote time zone database.

There is an update mechanism in libc++, however this is not done automatically. Invoking the function std::chrono::remote_version() will parse the version information of the tzdata.zi file and return that information. Similarly, std::chrono::reload_tzdb() will parse the tzdata.zi and leap-seconds.list again. This makes it possible to update the database if needed by the application and gives the user full power over the update policy.

This approach has several advantages:

  • It is simple to implement.

  • The library does not need to start a periodic background process to poll changes to the filesystem. When using a background process, it may become active when the application is busy with its core task, taking away resources from that task.

  • If there is no threading available this polling becomes more involved. For example, query the file every x calls to std::chrono::get_tzdb(). This mean calls to std::chrono::get_tzdb() would have different performance characteristics.

The small drawback is:

  • On platforms with threading enabled updating the database may take longer. On these platforms the remote database could have been loaded in a background process.

Another issue with the automatic update is that it may not be considered Standard compliant, since the Standard uses the wording “This update shall not impact the program in any way”. Using resources could be considered as impacting the program.