Lovely. Here's a list of all the charsets Mantis languages use:

big5 euc-jp euc-kr gbk iso-8859-1 iso-8859-15 iso-8859-2 iso-8859-9 koi8-r shift_jis utf-8 windows-1251 windows-1252 windows-1257

This list was produced with the following command, which is a mess but works:

cat lang/*.txt | grep charset | awk '{print $3}' | sed 's/['\''""; ]//g' | tr "[:upper:]" "[:lower:]" | sort | uniq

The list of supported charsets for htmlspecialchars() is shown here:

http://www.php.net/htmlspecialchars

The charsets Mantis uses which are not on that list are as follows:

euc-kr gbk iso-8859-2 iso-8859-9 windows-1257

I imagine iso-8859-2, iso-8859-9 and windows-1257 can harmlessly be handled as iso-8859-1 but I have no idea about euc-kr or gbk.

Problem disappear if you translate string_*.txt files with unsupported code pages to UTF-8, and of course set variable "$s_charset" in this files to 'UTF-8'.

UTF-8 is supported by htmlspecialchars and contains chars from unsupported code pages like iso-8859-2 for example.

Without a doubt. Conversion of Mantis to UTF-8, however, is a whole 'nother ball game; see 0004084.