Bug 170369
| Summary: | Input events precede keydown/keyup in Sogou IME | ||
|---|---|---|---|
| Product: | WebKit | Reporter: | Beth Dakin <bdakin> |
| Component: | HTML Editing | Assignee: | Nobody <webkit-unassigned> |
| Status: | NEW | ||
| Severity: | Normal | CC: | ap, bdakin, charleyroy, dbates, rniwa, ssaviano, webkit-bug-importer, webkit-unassigned, wenson_hsieh |
| Priority: | P2 | Keywords: | InRadar |
| Version: | Safari 10 | ||
| Hardware: | Unspecified | ||
| OS: | Unspecified | ||
| See Also: | https://bugs.webkit.org/show_bug.cgi?id=170368 | ||
| Bug Depends on: | 169209, 170370 | ||
| Bug Blocks: | 170368 | ||
Beth Dakin
This clone represents this action item: Input events precede keydown/keyup in Sogou IME
+++ This bug was initially created as a clone of Bug #169209 +++
I tested this on:
* Safari Technology Preview Release 23 (Safari 10.2, WebKit 12604.1.5)
* Safari Version 10.0.3 (12602.4.8)
I am using the Sogou IME (http://pinyin.sogou.com/mac/).
I am using this JS fiddle to log events, but any event logger will show this: https://jsfiddle.net/c1g6rahm/2/
In a content editable, press a punctuation character (eg. the key '.' to get '。', ',' to get '，', 'Shift+1' to get '！' - notice they are full width).
Obtained event trace:
1. textInput
2. input
3. keydown - 229 (WIN_IME key)
4. keyup - 188 (in this case I pressed ',')
This is problematic for a number of reasons:
1. There is no compositionstart/compositionupdate/composition end events, which there should be when using keydown WIN_IME
2. The textInput/input events are not between keydown/keyup, which they should be (they are when pressing regular ASCII/latin characters)
3. There is no keypress event fired. For example, Chrome fires a keypress with keycode 65292 for '，' and with keycode 12290 for '。'
The combination of these make it very hard to detect that these specific characters were selected without relying on the contents of the contenteditable. This also diverges from the spec at https://w3c.github.io/uievents/#events-composition-key-events.
| Attachments | ||
|---|---|---|
| Add attachment proposed patch, testcase, etc. |
Alexey Proskuryakov
This is a historical consequence of matching MSIE behavior for IMEs. It is kind of intentional, but so weird that it's probably not very important.
Is there a formal spec for inline input these days, which reasonable degree of cross-browser compatibility?