107.3 Lesson 1

Time zones are roughly proportional discrete bands of Earth’s surface spanning the equivalent to one hour, that is, regions of the world experiencing the same hour of the day at any given time. Since there is not a single longitude that can be considered as the beginning of the day for the whole world, time zones are relative to the prime meridian, where the Earth’s longitude angle is defined to be 0. The time at the prime meridian is called the Coordinated Universal Time, by convention abbreviated to UTC. Due to practical reasons, time zones do not follow the exact longitudinal distance from the reference point (the prime meridian). Instead, time zones are artificially adapted to follow the borders of countries or other significant subdivisions.

The political subdivisions are so relevant that time zones are named after some major geographical agent in that particular area, usually based on the name of a large country or city inside the zone. Yet, time zones are divided according to their time offset relative to UTC and this offset can also be used to indicate the zone in question. The time zone GMT-5, for example, indicates a region for which UTC time is five hours ahead, i.e. that region is 5 hours behind UTC. Likewise, the time zone GMT+3 indicates a region for which UTC time is three hours behind. The term GMT — from Greenwich Mean Time — is used as a synonym for UTC in the offset based zone names.

A connected machine can be accessed from different parts of the world, so it is a good practice to set the hardware clock to UTC (the GMT+0 time zone) and leave the choice of the time zone to each particular case. Cloud services, for example, are commonly configured to use UTC, as it can help to mitigate occasional inconsistencies between local time and time at clients or at other servers. In contrast, users who open a remote session on the server may want to use their local time zone. Thus, it will be up to the operating system to set up the correct time zone according to each case.

In addition to the current date and time, command date will also print the currently configured time zone:

The offset relative to UTC is given by the -03 value, meaning that the displayed time is three hours less than UTC. Therefore, UTC time is three hours ahead, making GMT-3 the corresponding time zone for the given time. Command timedatectl, which is available in distributions using systemd, shows more details about the system time and date:

As shown in the Time zone entry, time zone names based on localities — like America/Sao_Paulo — are also accepted. The default time zone for the system is kept in the file /etc/timezone, either by the zone’s full descriptive name or offset. Generic time zone names given by the UTC offset must include Etc as the first part of the name. So, to set the default time zone to GMT+3, the name of the time zone must be Etc/GMT+3:

Although time zone names based on localities do not require the time offset to work, they are not so straight forward to choose from. The same zone can have more than one name, which can make it difficult to remember. In order to ease this issue, command tzselect offers an interactive method which will guide the user towards the correct time zone definition. Command tzselect should be available by default in all Linux distributions, as it is provided by the package that contains necessary utility programs related to the GNU C Library.

Command tzselect will be useful, for example, for a user who wants to identify the time zone for “São Paulo City” in “Brazil”. tzselect starts by asking the macro region of the desired location:

Option 2 is for (North and South) American locations, not necessarily in the same time zone. It is also possible to specify the time zone with geographical coordinates or with the offset notation, also known as the Posix TZ format. The next step is to choose the country:

Brazil’s territory spans four time zones, so the country information alone is not enough to set the time zone. In the next step tzselect will require the user to specify the local region:

Not all locality names are available, but choosing the closest region will be good enough. The given information will then be used by tzselect to display the corresponding time zone:

The resulting time zone name, America/Sao_Paulo, can also be used as the content of the /etc/timezone to inform the default time zone for the system:

As stated by the output of tzselect, the environment variable TZ defines the time zone for the shell session, whatever the system’s default time zone is. Adding the line TZ='America/Sao_Paulo'; export TZ to the file ~/.profile will make America/Sao_Paulo the time zone for the user’s future sessions. The TZ variable can also be temporarily modified during the current session, in order to display the time in a different time zone:

In the example, command env will run the given command in a new sub-shell session with the same environment variables of the current session, except for the variable TZ, modified by the argument TZ='Africa/Cairo'.

Many regions adopt daylight savings time for part of the year — when clocks are typically adjusted by an hour — that could lead a misconfigured system to report the wrong time during that season of the year.

The /etc/localtime file contains the data used by the operating system to adjust its clock accordingly. Standard Linux systems have files for all time zones in the directory /usr/share/zoneinfo/, so /etc/localtime is just a symbolic link to the actual data file inside that directory. The files in /usr/share/zoneinfo/ are organized by the name of the corresponding time zone, so the data file for the America/Sao_Paulo time zone will be /usr/share/zoneinfo/America/Sao_Paulo

As the definitions for the daylight savings time may change, it is important to keep the files in /usr/share/zoneinfo/ up to date. The upgrade command of the package management tool provided by the distribution should update them every time a new version is available.

Linux systems can work with a wide variety of languages and non-western character encodings, definitions known as locales. The most basic locale configuration is the definition of the environment variable LANG, from which most shell programs identify the language to use.

The content of the LANG variable follows the format ab_CD, where ab is the language code and CD is the region code. The language code should follow the ISO-639 standard and the region code should follow the ISO-3166 standard. A system configured to use Brazilian Portuguese, for example, should have the LANG variable defined to pt_BR.UTF-8:

As seen in the sample output, the LANG variable also holds the character encoding intended for the system. ASCII, short for American Standard Code for Information Interchange, was the first widely used character encoding standard for electronic communication. However, since ASCII has a very limited range of available numerical values and it was based on the English alphabet, it does not contain characters used by other languages or an extended set of non-alphabetical symbols. The UTF-8 encoding is a Unicode Standard for the ordinary western characters, plus many other non-conventional symbols. As stated by the Unicode Consortium, the maintainer of the Unicode Standard, it should be adopted by default to ensure compatibility between computer platforms:

Some systems may still use ISO defined standards — like the ISO-8859-1 standard — for the encoding of non-ASCII characters. However, such character encoding standards should be deprecated in favor of Unicode encodings standards. Nevertheless, all major operating systems tend to adopt the Unicode standard by default.

System wide locale settings are configured in the file /etc/locale.conf. Variable LANG and other locale related variables are assigned in this file like an ordinary shell variable, for example:

Users can use a custom locale configuration by redefining the LANG environment variable. It can be done for the current session only or for future sessions, by adding the new definition to the user’s Bash profile in ~/.bash_profile or ~/.profile. Until the user logs in, however, the default system locale will still be used by user independent programs, like the display manager’s login screen.

In addition to the LANG variable, other environment variables have an affect on specific locale aspects, like which currency symbol to use or the correct thousand separator for numbers:

The locale command will show all defined variables in the current locale configuration:

The only undefined variable is LC_ALL, which can be used to temporarily override all the other locale settings. The following example shows how the date command — running in a system configured to pt_BR.UTF-8 locale —  will modify its output to comply with the new LC_ALL variable:

The modification of the LC_ALL variable made both abbreviations for the day of the week and month name to be shown in American English (en_US). It is not mandatory, however, to set the same locale for all variables. It is possible, for example, to have the language defined to pt_BR and the numerical format (LC_NUMERIC) set to the American standard.

Some localisation settings change how programs deal with alphabetical ordering and number formats. Whilst conventional programs are usually prepared to correctly choose a common locale for such situations, scripts can behave unexpectedly when trying to correctly alphabetically order a list of items, for example. For this reason, is is recommended to set the environment variable LANG to the common C locale, as in LANG=C, so the script produces unambiguous results, regardless the localisation definitions used in the system where it is executed. The C locale only conducts a simple bytewise comparison, so it will also perform better than the others.

Text may appear with unintelligible characters when displayed on a system with a character encoding configuration other than the system where the text was created. The command iconv can be used to solve this issue, by converting the file from its original character encoding to the desired one. For example, to convert a file named original.txt from the ISO-8859-1 encoding to the file named converted.txt using UTF-8 encoding, the following command can be used:

The option -f ISO-8859-1 (or --from-code=ISO-8859-1) sets the encoding of the original file and option -t UTF-8 (or --to-code=UTF-8) sets the encoding for the converted file. All encoding supported by command iconv are listed with command iconv -l or iconv --list. Instead of using the output redirection, like in the example, option -o converted.txt or --output converted.txt could also be used.