I'm not sure whether this belongs to the "RichFaces Users" or "RichFaces Developers" but...
I've been implementing a suggestionbox component recently, and realized that it behaves strangely in my Japanese environment here. To input Japanese text, I generally use an IME (Input Method Editor), which preprocesses the keystrokes to input a particular character. The suggestionbox apparently misses this and misfires on the first character.
I'm currently developing on:
Windows XP Pro (Japanese)
JBoss Portal Server 2.6.4
JBoss Portlet Bridge 1.0.0B2
For example, trying to input the character "ko" (u3053) into the suggestionbox I type 'k' then 'o' then hit the 'enter' key to accept this _one_ character, but the suggestionbox opens on 'k' and then does not fire again until after the IME accepts the character. This means that I can input a matching character u3053 that is not recognized by the suggestion method (which only checks 'k' and ignores u3053).
I've never dealt programmatically with the IME before, but a quick search indicates that javax.swing.text.AttributeSet has an attribute "StyleConstants.ComposedTextAttribute)" which shows when the character is fully composed from the IME. Perhaps this would help in fixing the problem. I can help with testing if any developer would like to look into it.
[edit: apparently this forum doesn't like Japanese either. Ick]