- From: Wenson Hsieh <wenson_hsieh@apple.com>
- Date: Thu, 11 May 2023 09:05:03 -0700
- To: public-editing-tf@w3.org
- Message-id: <9A665FD1-9948-4FD9-9B20-858B4C7C1E5A@apple.com>
[08:03] <whsieh> johanneswilm: start with issue 427, TPAC participation [08:04] <johanneswilm> https://github.com/w3c/editing/issues/427 [08:05] <whsieh> johanneswilm: let's do what we did in TPAC 2022 [08:05] <johanneswilm> https://github.com/w3c/editing/issues/425 [08:06] <whsieh> johanneswilm: next: https://github.com/w3c/editing/issues/425 [08:08] <whsieh> snianu: does not seem to be a way to deliver data async [08:08] <whsieh> snianu: (on macOS) [08:10] <whsieh> evan: if photoshop on Mac doesn't have this ability, they probably wouldn't ask for it on web? [08:10] <whsieh> johanneswilm: have we asked about macOS native app? [08:10] <whsieh> snianu: let's ask them what they do on macOS [08:11] <whsieh> whsieh: I can ask our own system framework folks for advice [08:11] <whsieh> https://github.com/w3c/editing/issues/424 [08:12] <whsieh> snianu: presented all options with partners: https://docs.google.com/document/d/1H6ow7RWa4MeycKP3OQBoMhLuMaJct3jSCWW6apcVu9U/edit [08:13] <whsieh> snianu: opt. 1 — add beforeunload event listener [08:14] <whsieh> opt. 2 — trigger all pasteboard callbacks before it fires beforeunload [08:14] <whsieh> opt. 3 — (most desirable) follow behavior of most native apps on windows: [08:14] <whsieh> ask the user if they want to preserve data on clipboard [08:15] <whsieh> address navigation performance impact by adding a timeout [08:19] <johanneswilm> johanneswilm: will copying few cells also lead to "large amount of data in clipboard" message? [08:19] <johanneswilm> snianu: yes, but app can choose not to do delayed rendering if very little is copied [08:19] <whsieh> evan: doing anything in beforeunload event would eject page from BF caceh [08:19] <whsieh> *cache [08:20] <whsieh> snianu: option 3 — not about beforeunload, just *before* unload? [08:21] == alexk [~alexk@a11e3e9c.public.cloak] has joined #editing [08:21] <whsieh> evan:showing dialog on navigation away would affect BFcache [08:22] <whsieh> snianu: it's opt-in so a website would be knowingly opting out of BFcache? [08:22] <whsieh> evan: beforeunload strongly discouraged, we should avoid adding new incentives to use this [08:23] <whsieh> evan: (the new incentive being, "delayed clipboard support") [08:23] <whsieh> snianu: what about intersection observers? [08:24] <whsieh> evan: BF cache is about "freeze-drying" webpage for quick loading later [08:25] <whsieh> evan: with beforeunload, never an opportunity here for a website to do something before it gets cached [08:25] <whsieh> snianu: let's discuss async [08:27] <whsieh> whsieh: motivation is to limit CPU usage? [08:27] <whsieh> evan: it's also to limit server workload [08:29] <whsieh> Chris: for the page unload case, is opt. 4 viable? [08:30] <whsieh> snianu: opt. 4 is undesirable from excel's standpoint [08:32] <whsieh> chris: wouldn't that only occur if the async clipboard delivery didn't occur in time? [08:32] <whsieh> chris: seems like an unlikely scenario [08:32] <whsieh> chris: should only be a second or 2 [08:32] == snianu [~snianu@a11e3e9c.public.cloak] has quit [Ping timeout: 180 seconds] [08:32] == No such nick or channel name: snianu [08:32] == End of WHOIS [08:33] == snianu [~snianu@a11e3e9c.public.cloak] has joined #editing [08:34] <whsieh> johanneswilm: the argument against putting all of them on the clipboard is that you're doing unnecessary work [08:34] <whsieh> johanneswilm: can take a few seconds for a document of 15 pages [08:35] <whsieh> chris: snianu, do you have data on time and memory it takes for this? [08:35] <whsieh> whsieh: (+1) [08:35] <whsieh> snianu: excel folks are working on it, we'll share async [08:35] <whsieh> chris: is 1 second ballpark? [08:35] <whsieh> snianu: unsure, let's double check... [08:35] <whsieh> snianu: probably more than a second [08:36] <whsieh> johanneswilm: from a user perspective, users who are used to copying, closing, pasting elsewhere would experience flakiness [08:37] <whsieh> johanneswilm: importantly user cannot understand why this is happening [08:37] <whsieh> johanneswilm: online Swedish banks ask users to copy/paste from apps [08:39] <whsieh> snianu: let's comment in issue 424 [08:39] <whsieh> snianu: will ask partners for concrete data on pasteboard delivery time [08:39] <whsieh> issue 19, EditContext [08:40] <whsieh> https://github.com/w3c/edit-context/issues/19 [08:40] <whsieh> dandclark: anne raises a good point, we should formalize element list [08:40] <whsieh> dandclark: use prior art from existing list? [08:40] <whsieh> dandclark: bare minimum: support div and canvas [08:41] <whsieh> dandclark: we do not want to be in the business of supporting EC on iframe, form controls [08:41] <whsieh> dandclark: proposal: list of elements that work with attached shadow [08:42] <whsieh> dandclark: "elements that are kind of boring" [08:42] <whsieh> dandclark: just need canvas [08:43] <whsieh> johanneswilm: list looks fine. on the first comment — would add span alongside div [08:44] <whsieh> (the list is https://github.com/w3c/edit-context/issues/19#issuecomment-1541033272) [08:44] <whsieh> chris: let's use the list of attached shadow elements + canvas [08:44] <whsieh> johanneswilm: let's resolve on that [08:44] <whsieh> https://github.com/w3c/edit-context/issues/36 [08:45] <whsieh> dandclark: EC — authors can supply geometry for where active composition, etc. are [08:46] <whsieh> dandclark: authors are expected to call updateCharacterBounds in characterboundsupdate event [08:46] <whsieh> dandclark: ambiguous — when is the author supposed to call that? [08:46] <whsieh> dandclark: feedback from adobe — they want to do the layout asynchronously [08:46] <whsieh> dandclark: works fine on windows using TSF [08:47] <whsieh> dandclark: on macOS, -[NSTextInputContext invalidateCharacterCoordinates] seems to be broken — IME window does not redraw until composition ends [08:48] <whsieh> dandclark: filed https://openradar.appspot.com/FB8936158 (rdar://72204786) [08:48] <whsieh> dandclark: we may not be able to implement it properly on macOS [08:49] <whsieh> dandclark: want to resolve — triggering async is okay [08:49] <whsieh> dandclark: robust w.r.t. async/multiprocess archtectures [08:51] <whsieh> dandclark: reduced repro — in any app on macOS, starting a composition in IME and scrolling [08:51] <whsieh> chris: https://bugs.chromium.org/p/chromium/issues/detail?id=1151301 [08:55] <whsieh> dandclark: TSF also doesn't know about scrolling [08:56] <whsieh> chris: invalidateCharacterCoordinates does not seem to do anything [08:58] <whsieh> whsieh: yeah let's resolve on async for now [08:58] <whsieh> snianu: https://chromium-review.googlesource.com/c/chromium/src/+/2586189 [08:59] <whsieh> chris: [meta] can we update calendar invite to include all repos? [08:59] <whsieh> johanneswilm: send an email with the calendar invite and let's fix it
Received on Thursday, 11 May 2023 16:05:37 UTC