Selecting locale at link time

RealView^® Compilation Tools for µVision Libraries and Floating Point Support Guide

Version 3.1

Home > The C and C++ Libraries > Tailoring locale and CTYPE using assembler macros > Selecting locale at link time

2.7.1. Selecting locale at link time

The locale subsystem of the C library can be selected at link time or extended to be selectable at runtime. The following describes the use of locale categories by the library:

The default implementation of each locale category is for the C locale. The library also provides an alternative, ISO8859‑1 (Latin‑1 alphabet) implementation of each locale category that you can select at link time.
Both the C and ISO8859‑1 default implementations usually provide one locale for each category to select at runtime.
You can replace each locale category individually.
You can include as many locales in each category as you choose and you can name your locales as you choose.
Each locale category uses one word in the private static data of the library.
The locale category data is read‑only and position independent.
scanf() forces the inclusion of the LC_CTYPE locale category, but in either of the default locales this adds only 260 bytes of read‑only data to several kilobytes of code.

ISO8859‑1 Implementation

Table 2.7 shows the ISO8859‑1 (Latin‑1 alphabet) locale categories.

Table 2.7. Default ISO8859‑1 locales

Symbol	Description
`__use_iso8859_ctype`	Selects the ISO8859‑1 (Latin‑1) classification of characters. This is essentially 7‑bit ASCII, except that the character codes 160‑255 represent a selection of useful European punctuation characters, letters, and accented letters.
`__use_iso8859_collate`	Selects the `strcoll`/`strxfrm` collation table appropriate to the Latin‑1 alphabet. The default C locale does not require a collation table.
`__use_iso8859_monetary`	Selects the Sterling monetary category using Latin‑1 coding.
`__use_iso8859_numeric`	Selects separating thousands with commas in the printing of numeric values.
`__use_iso8859_locale`	Selects all the ISO8859‑1 selections described in this table.

There is no ISO8859‑1 version of the LC_TIME category.

Shift‑JIS and UTF‑8 Implementation

Table 2.8 shows the Shift‑JIS (Japanese characters) or UTF‑8 (Unicode characters) locale categories.

Table 2.8. Default Shift‑JIS and UTF‑8 locales

Function	Description
`__use_sjis_ctype`	Sets the character set to the Shift‑JIS multibyte encoding of Japanese characters
`__use_utf8_ctype`	Sets the character set to the UTF‑8 multibyte encoding of all Unicode characters

The following describes the effects of Shift‑JIS encoding:

The ordinary ctype functions behave correctly on any byte value that is a self‑contained character in Shift‑JIS. For example, half‑width katakana characters, which Shift‑JIS encodes as single bytes between 0xA6 and 0xDF, is treated as alphabetic by isalpha().
The multibyte conversion functions, such as mbrtowc(), mbsrtowcs(), and wcrtomb(), all convert between wide strings in Unicode and multibyte character strings in Shift‑JIS.
printf("%ls") converts a Unicode wide string into Shift‑JIS output, and scanf("%ls") converts Shift‑JIS input into a Unicode wide string.

You can arbitrarily switch between multibyte locales and single‑byte locales at runtime if you include more than one in your application. By default, only one locale at a time is included.

Copyright © 2007 ARM Limited. All rights reserved.	ARM DUI 0378A
PDF version

Home > The C and C++ Libraries > Tailoring locale and CTYPE using assembler macros > Selecting locale at link time