Version 2.1.31 (20040620)
=========================
If no bugs are found in this release, the next release (which
will be identical or nearly identical to the current release)
will be numbered 2.2.0, so a new stable branch of the `ejudge'
system will be started. The 2.1 branch will be terminated. All
the support work will go to 2.2 branch, and all the new
development will go to the future development branch 2.3.

Support for different charset encodings is significantly
improved. All the explicit references to the "KOI8-R" charset
encoding are removed from the source code and replaced with the
encoding, specified in the --enable-charset option of the
configure script. If this option was not specified, the "UTF-8"
charset encoding is used.

The encoding specified in the --enable-charset option of the
configure script is used as the "internal" encoding, i. e. the
encoding which is used to store varous character strings in
memory. Thus the internal encoding must be compatible with the
string functions of the C Standard Library, i.e. the internal
encoding must not allow NUL ('\0') bytes in the middle of
character strings. For example, UTF-16 is not such an encoding,
whereas UTF-8 is.

During parsing of the XML configuration files and internal files
the external charset of the XML document (specified in the
document header) is converted into the internal charset. The
`iconv' function of the C Standard Library is used, so almost
all reasonable charset encodings are supported. All the
generated XML documents are also written in the internal
charset.

When a HTML file is generated by the ejudge programs, its
encoding is specified as defined in the corresponding
configuration files (by default it is the internal encoding).
However the current version of the ejudge system does not
support converting strings in the internal charset into the
charset of the HTML document, if they are different. The strings
are inserted into the HTML without conversion, which may result
in incorrent HTML documents. Thus, in the current version of the
ejudge system for correctness and safety the internal charset
and the output HTML document charset must be the same. So it is
not recommended to explicitly specify the charset encoding for
the generated HTML documents in the ejudge configuration files.

serve
-----
The `standings_charset' global configuration variable is no
longer supported. The implementation of its support was
incorrect and insecure in the previous versions. Is it possible,
that when charset conversion from the internal enconding to the
HTML document encoding will be supported, this configuration
variable will be restored.

ejudge-setup
------------
Several bugs in the generation of the ejudge setup script are
fixed.