GT.M supports international character sets using Unicode. A mumps process can operate in one of two modes: M mode and UTF-8 mode, which is specified by the environment variable gtm_chset at process startup and which is immutable for the life of the process.

In M mode, the process interprets strings as single byte characters with characters from $Char(0) through $Char(127) encoded as ASCII (M mode). A GT.M process in M mode places no interpretation on characters $Char(128) through $Char (255) – the application is free to use any interpretation it chooses, and ISO-8859 variants are commonly used. There is no distinction between a string of bytes and a string of characters and all bytes are valid characters – the concept of an illegal or a non-canonical character is a non-sequitur in M mode.

In UTF-8 mode, the process by default interprets strings as multi-byte characters encoded using UTF-8 (UTF-8 mode). Some sequences of bytes are not defined by the standard, and are considered illegal characters. The process can also interpret a string as a sequence of bytes, and the same string can have different properties (such as its length) when it is considered a sequence of bytes than as a sequence of characters.

Except when triggers are used, this interpretation is at the level of the process: the database treats strings as binary data and does not care how they are encoded. So, it is possible for the same database to be concurrently accessed by mumps processes operating in both M mode and UTF-8 mode. Mupip and other processes are not concerned with this distinction. If the database has triggers, since triggers are compiled code, all proceses that use a database must have the same mode.

The mode of a process is controlled by the environment variable gtm_chset. If it is not set, or has a value other than "UTF-8", the process operates in M mode. Within a process the ISV $ZCHset can be used to test the mode.

For a process to operate in UTF-8 mode requires:

ICU with a level of 3.6 or higher packaged as libicu. GT.M may also require the environment variable gtm_icu_version to be defined (the gtmprofile script attempts to detect and set it, if it is not set, but is not guaranteed to succeed).

The environment variable LC_CTYPE (or the environment variable LC_ALL) to specify a UTF-8 locale available on the system.

The locale command provides a list of available locales:

gtmuser@gtmworkshop7:~$ locale -a
C
POSIX
en_US.utf8
gtmuser@gtmworkshop7:~$
You can use the localedef command to generate additional UTF-8 locales – in this example, a UTF-8 locale for Arabic as used in Jordan is generated.

gtmuser@gtmworkshop7:~$ sudo localedef -f UTF-8 -i ar_JO ar_JO.utf8
gtmuser@gtmworkshop7:~$ locale -a
C
POSIX
ar_JO.utf8
en_US.utf8
gtmuser@gtmworkshop7:~$