About Unicode

Unicode is a 16-bit character set, portable across all major computing platforms and with decent coverage of almost all of the world. It is also single-locale; it includes no code pages or other complexities that make software harder to write and test. Finally, there is nothing else that's reasonably cross-platform. For these reasons, Trolltech has chosen to make Unicode the native character set of Qt starting with version 2.0.

Information about Unicode on the web. The Unicode Consortium has a number of documents available, including

• A technical introduction to Unicode
• The home page for the standard

The Standard. The current version of the standard is 3.0.0.

• The Unicode Standard, version 2.0. See also the 2.1 update and the 2.1.9 data files at www.unicode.org.
• The Unicode Standard, version 3.0. See also its home page.

As used in Qt. In Qt, and in most applications that use Qt, most or all user-visible strings are stored in Unicode, and Qt provides

• Translation to/from legacy encodings for file I/O - see QTextCodec and QTextStream
• Translation from Input Methods and 8-bit keyboard input
• Translation to legacy character sets for on-screen display
• A string class, QString, that stores Unicode characters, with support for migrating from C strings including fast (cached) translation to and from the US-ASCII, and all the usual string operations
• Unicode-aware widgets where necessary
• On Windows 95/98/NT/2000, Unicode support detection, so that Qt provides Unicode even on Windows platforms that do not support it

To obtain the benefits of Unicode, we recommend using QString for storing all user-visible strings and do all text file I/O using QTextStream. Use QKeyEvent::text() for keyboard input in any custom widgets you write; it does not make much difference for slow typists in West Europe or North America, but for fast typists or people using special input methods using text() is beneficial.

All the function arguments in Qt that may be user-visible strings, QLabel::setText() and a thousand others, take const QString & as type. QString provides implicit casting from const char * such that things like

        myLabel->setText( "Hello, Dolly!" );

will work. There is also a function, QObject::tr(), that provides translation support, like this:

        myLabel->setText( tr("Hello, Dolly!") );

tr(), oversimplifying a bit, maps from const char * to a Unicode string, and uses installable QTranslator objects to do the mapping.

Turning back to Unicode, for programs that needs to talk to other programs or read/write files in legacy file formats, Qt provides a number of built-in QTextCodec classes, that is, classes that know how to translate between Unicode and a legacy encoding.

By default, conversion to/from const char * uses a locale-dependent codec. However, the program can easily find codecs for other locales, and set any open file or network connection to use a special codec. It is also possible to install new codecs, for encodings that the built-in ones do not support. (At the time of writing, Vietnamese/VISCII is one example of that.)

Since US-ASCII and ISO-8859-1 are so common, there are also specially fast functions for mapping to and from them. For example, to open an application's icon one might do this:

        QFile f( QString::fromLatin1("appicon.png") );

Regarding output, Qt will do a best-effort conversion from Unicode to whatever encoding the system and fonts provide. Depending on operating system, locale, font availability and Qt's support for the characters used, this conversion may be good or bad. We aim to extend this in upcoming versions, with emphasis on the most common locales first.