Browser-floor cleanup: drop legacy polyfills + modernise wheel/scroll

[moodyjmz]

Low-hanging-fruit browser-floor cleanup — removes dead weight, no behavioural risk on the modern floor.

• 🔥 Stop deploying fetch / es6-promise polyfills (unused once IE/legacy is off the floor).
• 🐛 Replace DOMMouseScroll / mousewheel UA-sniffing with the standard wheel event across desktop editors.

Details

Polyfills — fetch and es6-promise are dropped from build/vendor.manifest.mjs, so the deploy step stops copying them. The vendor dirs stay in the tree (upstream keeps shipping them); we just don't deploy them.

Wheel/scroll — three fixes in the desktop editors' Print.js / DocumentHolder.js:

1. Replace the DOMMouseScroll/mousewheel UA-sniff with the standard wheel event.
2. Switch Print.js binding from jQuery .on() to addEventListener.
3. Correct the wheel delta sign in Print.js and DocumentHolder.js.

Part of the browser-floor-stage1 work; the mobile/ES2022 half is a separate PR.

[MonaAghili]

as far as I understand $previewBox is a jquery obj and in current version of it which is 3.5.1 the wheel event doesn't expose deltaY on the jquery event object which means e.deltaY would be undefined so forward wil lbe always false

would it make sense to change the event back to mousewheel in all printjs files? The jquery.mousewheel plugin can provide e.deltaY for mousewheel events

[moodyjmz]

Good catch — you're right. jQuery 3 wraps events and doesn't copy deltaY into the wrapper, so e.deltaY was undefined and forward always evaluated to false.

Fixed in 230a052: switched all five Print.js files from $previewBox.on('wheel', ...) to native $previewBox[0].addEventListener('wheel', ..., {passive: false}). The {passive: false} is needed because onPreviewWheel calls preventDefault() on Ctrl+scroll — consistent with how DocumentHolder.js already binds its wheel listener.

[moodyjmz]

Fair point, and you were right about the version you saw — jQuery wasn't exposing deltaY on the wheel event. Since switched to native addEventListener('wheel', …) which does. Apologies for the drive-by review request though — Claude got a bit eager and pinged reviewers when this should've gone up as a draft first. Marking it draft now while I finish the runtime testing. :P

[MonaAghili]

@moodyjmz
No need to apologize!!! we're a team and the whole point of reviews is to catch things together and make the project better
It happens to every single of us especially in a big codebase even claude misses things sometimes!

other than that I didn't find any other issues and LGTM!

[moodyjmz]

Independent review caught two sign-inversion bugs introduced by this PR — both now fixed in 0ffa044.

---

Root cause

The old jQuery .on('mousewheel', handler) path routed through vendor/perfect-scrollbar/src/jquery.mousewheel.js, which negates native deltaY to up-positive (scroll up = positive, matching legacy wheelDelta). Switching to native addEventListener('wheel') bypasses that plugin entirely. TabBar.js._onMouseWheelThrottled was already written to account for this (-deltaY). The other two handler families were not.

Bug 1 — Print preview scroll direction inverted (all browsers)

onPreviewWheel in all five Print.js files:

// before: down-scroll → deltaY > 0 → forward = false → previous page (wrong)
var forward = (e.deltaY || ...) < 0;
// after
var forward = e.deltaY > 0;

This regressed Chrome, which worked before the PR.

Bug 2 — Ctrl+scroll zoom direction inverted (Firefox)

handleDocumentWheel / onDocumentWheel in all five editors:

// before: Firefox has no wheelDelta → fell through to raw deltaY → down-positive
//         scroll up → deltaY < 0 → delta < 0 → zoomOut (wrong)
var delta = (_.isUndefined(event.originalEvent)) ? event.wheelDelta : event.originalEvent.wheelDelta;
if (_.isUndefined(delta)) { delta = event.deltaY; }

// after: negate deltaY → up-positive → existing delta>0=zoomIn stays correct
var delta = event.deltaY !== undefined ? -event.deltaY : event.wheelDelta;

Firefox had no working Ctrl+zoom before this PR (DOMMouseScroll path was dead). The fix makes it correct rather than merely swapping dead for inverted.