Recently i reported a problem of broken trackback by #1647.
But I think you have some misunderstanding.
[3081] and [3107] don't help this problem solved.
If we regard one capital and one small letter as one character,
let us assume that AaBbCcDdEeFf is excerpted by 10bytes.
Before : $excerpt is AaBbCcDdEeFf
in /wp-includes/functions-post.php do_trackback function
$excerpt = substr($excerpt, 0, 7) . '...';
After : $excerpt is AaBbCcD...
Then Dd is cut off and it becomes a broken character.
With another blog tool, it is printed like AaBbCc?..., because D is broken character.
But in case of wordpress,
in wp-trackback.php
if ( function_exists('mb_convert_encoding') ) { // For international trackbacks
$title = mb_convert_encoding($title, get_settings('blog_charset'), $charset);
$excerpt = mb_convert_encoding($excerpt, get_settings('blog_charset'), $charset);
$blog_name = mb_convert_encoding($blog_name, get_settings('blog_charset'), $charset);
}
mb_convert_encoding function considers that AaBbCcD... is not UTF-8 because D is broken.
Therefore AaBbCcD... is coverted UTF-8 by other endodings, and every character is broken after all.
Wordpress uses mbstring module for international trackback. So open /wp-includes/functions-post.php and find do_trackbacks function
$excerpt = substr($excerpt, 0, 252) . '...';
replace it with
if ( function_exists('mb_strcut') ) // For international trackbacks
$excerpt = mb_strcut($excerpt, 0, 252, get_settings('blog_charset')) . '...';
else $excerpt = substr($excerpt, 0, 252) . '...';
Must use mb_strcut Not mb_substr!
Because mb_substr is cut by character and mb_strcut is cut by byte.
for example
mb_strcut('AaBbCc', 1, 2) returns 'Aa'.
Treated as byte stream.
mb_substr('AaBbCc', 1, 2) returns 'BbCc'
Treated as character stream.
