Meeting minutes
Multi-pen support and persistent pointerId #353 w3c/pointerevents#353
related PR w3c/
Rob: I left my +1 to the "sure, let's call it persistentDeviceId and put it on the event"
Olli: I also agree, not sure what default values would be
Rob: for default value, it could be 0 for developers to do truthiness checks
Rob: we were originally going to have 0 as default, but some implementation used 0 for the mouse
Olli: initially there wasn't any decision, some used 0, some 1 ...
Rob: suggest as there's no precedent, we use 0 for uninitialised or unknown
Olli: would be nice if the id was random...
Rob: we're not reserving any id values for this, and it's supposed to be random
Olli: need to make sure that implementations don't just default to something like 1 for mouse
Olli: could do a WPT to check that persistentDeviceId for a mouse is different - open two origins/documents and verify that they don't have the same id for mouse
Olli: test could be cross-origin
Rob: sounds reasonable
[discussion around whether or not different OSs support multiple mouse pointers - all OSs pretend it's one device?]
Rob: I believe Chrome does give a persistent id for mouse, so we can test that
Patrick: so for next steps - I will explore setting up a new vNext branch, retarget this PR to that, then we should do one final review on this and can then merge it into that. Will follow up again with Philippe about whetehr or not new branch will cause problems
Olli: we should also work on test now, so we don't have to circle back
ACTION: Patrick to set up new vNext branch and repoint PR there, group to review PR
Meta-issue: update WPT to cover Pointer Events Level 3 #445 w3c/pointerevents#445
Patrick: did you all manage to meet up for some WPT work two weeks ago?
mustaq: yes, we were able to close a few of the outstanding WPTs, should get there in a few more days
Olli: have a few more to do, but made good progress
Rob: I didn't have a chance, but we're in a good spot - we know what we need/want to test
Olli: so we have 2 left now
<mustaq> Two more left: #477 and #300
<smaug> whatwg/
AOB
Olli: the above just came up in some other discussion
Olli: mozilla internal triage of spec issues
Rob: i believe, mustaq, that your comment says we can't always consider the down for non-mouse as activation, but there may be some mitigations.
Rob: we could distinguish based on the cancelability of pointerdown whether it gives activation
Rob: if you haven't changed the touch-action, cancelling the pointerdown does not prevent scrolling
Rob: i don't know how we represent that state, I believe it's a bit handwavy in the spec. but there IS a difference betwen pointerdowns that CAN cause scrolling/can't be cancelled, and once that don't. this could help situation a bit, even if we can't fully resolve it
Rob: as dev, i fear danger of developer writing a pointerDown listener, testing on desktop with mouse, and then not realising it's not triggering/activating on touch/stylus
mustaq: there is a fundamental difference between dragging with mouse vs dragging with finger...
mustaq: Olli if you can add more context to the issue of what challenges you face...
Olli: oh this was only part of our triage, so no immediate challenge
Olli: not seen any recent bug reports for this
Rob: my concern is you can't use pointerevents reliably for user activation
Olli: forces people to use something like click. yeah we can continue thinking about this, but just as a reminder
Rob: was hoping that we can establish if you're in a situation where the pointerdown can't result in a scroll...
Rob: or we just say you can't get activation until click event (even for mouse), but that could potentially break implementations
mustaq: i can't see a way to consolidate them, not all pointerdown will result in scroll, but the spec should say something either way
Patrick: purely from my luddite view, I quite like the idea to try and not cause "accidental activation", if there's a way to differentiate (using cancelability or similar) a pointerdown on touch/stylus that is a scroll vs one that is doing something (because the author has used touch-action), then that would make sense
[discussion that it's not directly the cancelability of the event that counts here, but some hint about whether it will or won't cause panning]
Patrick: thank you all, catch you again in 2 weeks' time