17:28:51 RRSAgent has joined #openui 17:28:51 logging to https://www.w3.org/2021/06/17-openui-irc 18:02:10 Francis_Storr has joined #openui 18:02:13 present+ 18:02:14 Zakim, start meeting 18:02:14 RRSAgent, make logs Public 18:02:15 please title this meeting ("meeting: ..."), gregwhitworth 18:02:21 present+ 18:02:24 present+ 18:02:26 present+ 18:02:26 present+ 18:02:36 meeting: Open UI Telecon 18:02:55 una has joined #openui 18:03:01 chair: Greg Whitworth 18:03:02 bkardell_ has joined #openui 18:03:03 davatron5000 has joined #openui 18:03:08 present+ 18:03:08 present+ 18:03:29 flackr has joined #openui 18:03:31 present+ 18:03:33 scribeNick: una 18:03:40 present+ 18:03:41 masonf has joined #openui 18:03:41 scribenick: una 18:03:45 andrico1234 has joined #openui 18:03:45 present+ 18:03:55 ScribeNick: una 18:04:31 present+ 18:04:42 gregwhitworth: How do we feel about having July 8th off given the US holiday 18:04:52 RESOLVED: We will not have an OpenUI meeting July 8th, 2021 18:05:18 gregwhitworth: Does OpenUI want to take on the incubation of anchor positioning with the expectation this will graduate to CSSWG? 18:05:40 melanierichards: We will still alert CSSWG we're working on this 18:06:04 una: i think theres enough here to have at least a few sessions incubating it 18:06:25 gregwhitworth: I think it would be good to frontload that due to the interest in this group and high priority for moving popup forward 18:06:44 melanierichards: It would be following week due to OOO next week 18:07:42 gregwhitworth: I meant July 1 instead of July 8th last time 18:07:57 Poll: 1 or 8 for which day you'd prefer off 18:07:59 1 18:08:01 Prefer 8th off 18:08:01 prefer 8th off 18:08:01 I prefer 1st off to 8th 18:08:02 8 18:08:04 0 18:08:05 8 18:08:07 1 or 8 fine (0) 18:08:08 8 18:08:46 5:2 for 8th to 1st, we'll have telcon on the 1st 18:09:32 gregwhitworth: let's make sure CSSWG is aware and can engage in this process 18:09:42 Topic: Should the search for parts pierce into shadow roots? 18:09:51 https://github.com/openui/open-ui/issues/347 18:10:22 dandclark: My examples here are about select menu but can apply to any customizable parts we want to build 18:10:47 dandclark: select menu is going to search its subtree to look for where to apply events and a11y elements 18:10:57 dandclark: I've referred to this at flat tree traversal 18:11:11 dandclark: shadow roots flattened out 18:11:28 dandclark: this mostly works for select menus to apply controller code to, but in some cases this dosnt make sense 18:11:50 dandclark: [shares screen to show some code samples] 18:12:10 ^ code examples are in the github issue 18:12:36 dandclark: if i nest selects options, with flat tree traversal the nested elements cause issues 18:13:05 dandclark: there should be another rule to this: a flat tree traversal doesn't descend into the subtree or shadow root of custom elements 18:13:21 q+ 18:13:22 q+ 18:13:24 dandclark: of another customizable element* or another select element 18:13:45 ack masonf 18:14:08 masonf: I put some opinions on this issue, on this particular item - i'd ask why any customizable element and not another select 18:14:14 masonf: you're looking for elements w/the 18:14:18 masonf: part attribute 18:14:32 masonf: if you made those names more unique, like select-part, you could. get around some of the issues 18:14:49 dandclark: theres a seperate issue about `part` being a problematic name 18:15:04 :thingr 18:15:07 dandclark: i assume when we switch to a diff name it would be the same for all of the customizable elements we want to biuld 18:15:15 dandclark: if its not the same, then I agree 18:15:44 masonf: that makes it a little more clear. if you had a common name for the new customizable control part, i see why youd have to stop at any other customizable control 18:16:04 dandclark: a customizable control shouldn't be lookign into another customizable tree -i thikn we agree 18:16:15 gregwhitworth: whats the definition of customizable control? 18:16:29 gregwhitworth: do you mean native shipping elements that follow this paradigm or web components in general? 18:16:40 dandclark: the next case i'm going to get into will talk abuot that 18:17:06 dandclark: i see it as a static list of tags 18:17:22 dandclark: set of known tags, i.e. select and selectmenu 18:17:39 gregwhitworth: i feel like this is where we'd just create a new definition in whatwg 18:18:19 q? 18:18:22 ack gregwhitworth 18:20:37 Proposed Resolution: In a customizable control's search for parts, the traversal should be a flat tree traversal, which does not descend into other customizable controls or their children, or the shadow root of a selectmenu, or a select. 18:21:13 ACTION: dandclark to define customizable controls 18:21:58 Proposed Resolution: In a customizable control's search for parts, the traversal should be a flat tree traversal, which does not descend into other customizable controls or their children, or user agent defined customizable controls. 18:22:31 Proposed Resolution: In a customizable control's search for parts, the traversal should be a flat tree traversal, which does not descend into other user-agent defined customizable controls or their children. 18:22:48 +1 18:22:48 RESOLVED: In a customizable control's search for parts, the traversal should be a flat tree traversal, which does not descend into other user-agent defined customizable controls or their children. 18:23:29 dandclark: for user-defined custom elements, what do we do about those? 18:24:35 dandclark: example: user defiend some custom element, has options they're slotting in from light DOM, and their shadow DOM is doing something 18:25:02 dandclark: should this select meny be applying controller code to these options 18:25:59 q+ 18:26:00 q+ 18:26:06 ack masonf 18:26:07 ack masonf 18:26:21 masonf: i think i am advocating for these options being seen by the select menu 18:26:35 masonf: partly depending on naming. if its a generic like `part`, thats problematic 18:26:45 masonf: if its customized then it opens the doors to possibility 18:26:49 ack bkardell_ 18:26:52 q+ 18:27:23 bkardell_: if we remove `` and hoist everything, do we agree that should work? 18:27:32 masonf: the question is should the shadow root hide anything 18:27:43 masonf: if you agree with. last resolution, it should work in that case 18:28:07 masonf: the custom element may have its own controller, which adds to he confounding here 18:28:15 s/masonf/bkardell_ 18:28:28 masonf: yes thats the concern 18:28:42 ack gregwhitworth 18:28:54 gregwhitworth: this is an odd usecase to me, not that it'll never happen 18:29:30 gregwhitworth: the second you say we're going to be applying our own behavior, i'd be intrigued in understanding that usecase 18:29:46 gregwhitworth: it becomes way more complicated, seems like a very rare scenario 18:30:13 dandclark: how would you distinguish if something has its own controls or not 18:31:16 gregwhitworth: when youre trying to add behavior into the behavior thats already there that the controller is trying to wire up, it gets really weird. my instinct is to prevent default in this use case and provide my own behavior 18:31:52 gregwhitworth: that usecase feels like the author would want to take on the ownership of that component 18:32:11 dandclark: masonf was saying its different if we use part attribute 18:32:25 dandclark: i..e. the select of today treats options by the tag 18:32:35 dandclark: thats how the current chromium prototype also works 18:32:49 dandclark: you're opting in to get this special controller code applied to you 18:34:07 masonf: it inadvertently makes it easier to apply controller code to customizable elements, the chance of this happening is lower with custom names 18:34:27 dandclark: we should figure this out (can controller code get applied to an option without a special attribute) 18:35:01 my only hesitance here is masonf's "the chances of ... seems low" just seems to historically be wildly underestimating the creativity of developers 18:35:09 gregwhitworth: I cant imagine introducing a new select menu and require each option to have some long code tied to it 18:35:50 gregwhitworth: aria roles that are appropriate for a select, when i cick on it - the list box click handler, keyboard handlers for focus and enter 18:35:59 s/gregwhitworth/dandclark 18:36:24 masonf: lets enumerate those on the issue, what we're looking for are breakages 18:37:13 masonf: what i want to avoid is a special new DOM traversal mechanism 18:37:43 gregwhitworth: to me, its: pretend the option element bc its baked into the UA, the select knows what an option is, so its going to treat that as controller code 18:38:13 masonf: I like/want that - the use of options without a fancy part 18:39:03 dandclark: what is breakage scenario if i have a select nested deep in another select menu 18:39:28 dandclark: itll obfuscate the collapsing/etc 18:39:58 gregwhitworth: issue is that theyre using the option naming, my thoughts are that this is a new element 18:40:11 gregwhitworth: lets detail the areas where this might break in the issue 18:40:23 dandclark: will anyone want to put a select in a select? 18:40:43 gregwhitworth: if they do, thye can change their `