Search suggestion doesn't support 2byte language such as CJK right now. It's different to detect finishing syllables. e.x onKeyUp and onKeyDown in JavaScript. So Google made 2 byte langunage version. http://www.google.co.jp/webhp?complete=1&hl=ja

I'm not sure what '2byte languages' are. Probably, all scripts that use IME/input methods for input have 'this issue' so that the summary is being changed accordingly. BTW, I'm afraid I'm too slow to get what attachment 222663 [details] tries to 'convey'. Channy, can you explain 'the issue' to me (off line)?

Is this still the same in a3?

> Is this still the same in a3? it doesn't work still. I will explain this problem to Jungshik.

Masayuki and Jungshik, do you guys know if there's an easy way to get at the Japanese IME input before it's committed to the text field in Firefox? I'd like to support Japanese Suggest specifically (which can use the IME input to give suggestions), but I don't know what APIs we have available. Unfortunately, there's no specific Google Suggest support for Corean or Chinese characters right now.

This function may make to happy Japanese people. But I don't think so Chinese and Korean people. E.g., MS's Korean Global IME commits per character. Is this breaking current suggestions by the composition string on Korean language? Jungshik: What do you think?

If it's possible as like http://www.google.co.jp/webhp?complete=1&hl=ja, korean character is nicely okay.

Y! seems to get this right, what's the difference in the implementation?

Joe, you told me a while back that you were looking into this, what's up?

My understanding is that we can support Japanese IME, if we can pass the partial IME input to the server (hence my question in comment #5, which I haven't gotten an answer to). Google doesn't have Korean or Chinese versions of Suggest, though.

Simon, can you answer comment 5?

I'm not at all familiar with our IME code, I'm afraid. If Masayuki and Jungshik aren't available right now, maybe Aaron or Kyle can help?

Unfortunately, I don't have much time for this...

(In reply to comment #10) > My understanding is that we can support Japanese IME, if we can pass the > partial IME input to the server Looks to me like typeaheadfind supports this by handling HandleText(nsIDOMEvent* aTextEvent), not to be confused with the handleText method defined by the autocomplete code.

--> unblocking. As far as we can tell, this isn't breaking Google search, and we've got no traction on why this is different from Y! suggest. The primary search provider (Y!) works fine, so while we'd take a well baked patch, we're not going to block on this issue.

Bug 354358 is a primary bug to solve it.