This is spun off from bug 433340. On mac, we need to notify the current window level to IME via kTSMDocumentWindowLevelPropertyTag. However, we don't notify it. And for doing it, we need to get the "actual" focused widget. The topmost <popup> element is not handling the key/IME events itself. Because it never has "native" focus. Therefore, we need to query the "actual" focused widget via NS_QUERY_TEXT_CONTENT event. Note that the kTSMDocumentWindowLevelPropertyTag should be set at every focus changed. However, the <popup> element never gets the native focus. Therefore, this patch set the property at every keydown event (only IME is enabled).

fix a mistake, sorry for the spam.

oops, s/<popup>/<panel>/ in comment 0.

I tested on following IMEs: on 10.5 * ATOK 2006 * Kotoeri * egbridge Universal 2 * ITABC * Zhuyin * Pinyin * Cangjie * Jianyi * Dayi(Pro) on 10.4 * ATOK 2006 * Kotoeri * egbridge Universal * ITABC * Hanin * Cangjie * Jianyi * Dayi(Pro) * Pinyin

Josh: Would you review this?

Steven? Would you review this?

Josh: Do still I need the Steven's review?

Sorry for the delay. I've been (and still am) on vacation. I'll get to this on Monday (8-11). Since the patch is a bit complicated, I feel obliged to do some manual testing.

Lets move ahead without Steven's review, but leave the review request there. I'd like to have him look even if it happens after something lands.

pushed to trunk.

Comment on attachment 330942 [details] [diff] [review] Patch v1.0.1 Looking through the code, I have only one comment: I suspect 'sDocumentID = nsnull;' shouldn't have been dropped from nsTSMManager::EndComposing(). But (as best I can tell) leaving this in or dropping it makes no difference in current code, so this is just a minor quibble. Otherwise everything looks fine. I tested (on OS X 10.5.4 and 10.4.11) with the Japanese Hiragana and Chinese Pinyin input methods, in combination with whatever oddball use of popup windows and menus (native and non-native) I could think of. I didn't see any problems.

Steven: Thank you for the review. I'll post the patch for the issue. For the safety, I agree your suggestion.