Meeting minutes
PE3 candidate recommendation https://www.w3.org/TR/pointerevents3/
Patrick: for those who haven't seen already, PE3 is now in CR. spoke to Philippe and asked about WPTs and whether we need everything passing in 2 implementations, or if it's on a per test basis.
Patrick: Need 2 implementations per test
Patrick: to then exit CR we need 2 implementations for each test. implementation report just pointing to WPT results
PLH: spec is now back on automatic publishing as well
Olli: luckily PE is part of interop 2025, so some of the tests/implementations will get fixed now
PointerEvents as living standard w3c/pointerevents#471 (comment)
[discussion on the challenges of living standard, and the editing overhead / how it makes specs unreadable, and other options]
Patrick: so any preference? I'm happy with NOT having too much editing overhead, happy to keep the regular model we've been doing
Rob: one thing I would suggest is clearly marking things that are at risk
PLH: that is also supported in bikeshed (?). but suggest we work on the tip, and then once we have 2 implementations for a set of new features, we cut that to a Level 4
Rob: one issue we also have is making changes to existing/previous issues
PLH: which is why, as you may have seen, we deprecated older versions
Mustaq: is there a way to version WPTs? to make clear when a test is for 2, or 3, ...
PLH: my recommendation is to always have WPT refer to the latest tip. not split by levels
PLH: if we need to, we can make a separate list of which WPTs are a sublist for a particular version
Olli: just checked MDN, it points to editors' draft
PLH: and editor draft is always the tip
PLH: and TR/pointerevents goes to the tip
PLH: versioned URLs become historical artifacts more than anything else
Patrick: is there a cutoff point now that we're in CR when we need all WPTs green?
PLH: in next 20 years ... but seriously, not maximum
Patrick: suggest we push to get these supported soon though, so we don't drag this on forever in CR
PLH: we may also need to see which WPTs we care for. example of touch-action with more than one value, seems Gecko doesn't support it, only blink
Olli: think there's also issues in WPT runners themselves
PLH: and that's ok. if we can come up with an explanation of why a test is red, that's ok
PLH: but yes, decide which WPTs do we care about, and then check if they fail because WPT is limited, or if there's a real lack of implementation, and if the latter, do we push for it
PLH: also, do we have full test coverage?
Olli: think so. Masayuki has found a lot of edge cases and filed new tests/implemented things in Gecko
Rob: WPT tests also have issues with coalesced etc. tests
PLH: so we may need to do manual test...
PLH: we need to compile a list of how tests *can* be run, do they need manual tests, are there some things that realistically *can't* even test
PLH: example of what may not be testable is how a pointerId gets assigned
Rob: there are also things that you need particular hardware for
Rob: such as detailed properties for pens
PLH: maybe then taking the list of changes, then check that each change (PE2 > PE3) has test, and note if there are gaps for testing (test needs to run manually, can't be tested, etc)
PLH: if we conclude that for some features we won't have sufficient implementation, we can then still remove them from level 3 and defer them to level 4
PLH: want me to make a first pass at this?
Patrick: if you have time/inclination, i'd very much appreciate it, thank you
PLH: might do separate markdown file and link from issue
ACTION: PLH to make first run at compiling list of new features, related WPTs, and any gaps in tests/implementations
Meta-issue: update WPT to cover Pointer Events Level 3 w3c/pointerevents#445 / https://github.com/w3c/pointerevents/issues?q=label%3Aneeds-wpt+
Patrick: unless I forgot to take labels off, it looks like we have 3 that still need wpt. how are they doing?
Adam: I worked on w3c/
Patrick: thank you, removing needs-wpt label
Patrick: so we have w3c/
Mustaq: i meant to do a test, will look at it
ACTION: Mustaq to look at test for #534
Patrick: and the last one w3c/
Olli: Masayuki was away last few days, will ask him if he can write a test
ACTION: Olli to ask Masayuki about test for #509
w3c/pointerevents#542
Mustaq: idea we had - if you can tie it to click event, identify if it's a stable click or a "drifting" click. from a web platform perspective
Patrick: that reminds me from old touch events tests i used to do, distinction between a "dirty" tap and a "clean"tap, and only the latter would also fire click
Mustaq: if you know of any other spec/issue that could be helped by this idea of drifting/dirty click, add to the issue
Mustaq: doing "slop detection" from basic principles, or if this could be helped via UA
Olli: need to be careful that we don't implement some magical property that then works differently based on different heuristics
TPAC 2025
Patrick: still early days, but start planning/thinking if you'll be in person at TPAC (november in Kobe, Japan). we have until 20 June to book slot for group session/joint sessions. For latter, was thinking perhaps trying to sit down with either i18n or csswg for the logical touch-action values. Anyway, not putting people on spot right now, but start considering
Olli: I will likely be there
Patrick: if there's no other topics, we'll keep it short-ish. Thank you all, we meet again in 2 weeks' time