SWI-Prolog is a Unicode application. This means all characters are internally encoded as integers (0..0x10ffff) in a uniform space. This encoding can represent any character in any language. Streams however are sequences of bytes, i.e., sequences of integers in the range 0..255. This implies some encoding is needed to represent Unicode characters. Various encodings exist, some of which cannot represent all Unicode characters. SWI-Prolog uses the default encoding of the platform on which it runs. If it must write a character that does not fit in this encoding it gives this error message.

There is no easy answer on what to do. Understanding international character encoding issues is hard. Luckily, more and more platforms now use UTF-8 as default encoding. If you need to deal with characters of other scripts than the default language we advice to check whether your OS can support UTF-8.

Regardless of the default locale, SWI-Prolog can be forced to use UTF-8 using

:- set_prolog_flag(encoding, utf8).

In addition, the encoding/1 directive may be used to declare the encoding of a source file regardless of the system encoding. This directive is supported by multiple Prolog systems. Some require it to be placed right at the beginning of the file (even comments are not allowed to precede the directive). In SWI-Prolog is may appear anywhere in the file and a single file may change the encoding multiple times.

See also

- Wide character support