Only the organizer may edit or move a meeting

[brhellman]

RFC 5546 section 2.1.4: only the organizer revises a meeting. The editor and dragging were gated on the calendar being writable alone, so an attendee could retitle, move or re-rule a meeting somebody else organizes. The write changed nothing but their own copy - every other guest still held the original - and a server implementing scheduling (Google does) is entitled to reject it outright. In practice: an attendee drags a meeting an hour later, sees it move, and is the only person in the world who thinks it did.

• CalendarEventPopover._isEditable() and canMoveEvent both refuse a meeting this account does not organize (isMine from Answer or counter an invitation from the calendar's right-click menu #2928 - the ORGANIZER is us, or the event names none and is nobody's meeting but ours, or Google rewrote the ORGANIZER to a ...@group.calendar.google.com id on a calendar the server (Record whether each calendar is the account's own or shared into it #2916) says is ours). A new event and a read-only calendar behave as before. The two paths share the test, so what can be dragged and what can be edited never disagree.
• A guest's drag is not silently ignored: it begins (canAttemptMove - writable calendar, not cancelled), and once it is clearly a drag the grid clears it and shows a dialog - "This meeting can't be rescheduled. Only the organizer can move it." - with Propose a new time and Cancel; the arrow keys get the same dialog. The button opens the counter-proposal picker the context menu uses.
• The read-only card an attendee lands on offers Propose a new time... in place of the pencil - the one affordance that replaces editing (RFC 5546 section 3.2.7, via Propose a new time for an invitation #2926's popover) - so it is not hidden behind a right-click.

Double-clicking an event we organize to open the editor directly is its own PR, stacked on this one.

Stacks on #2928 (isMine) and through it on the scheduling PRs; carries their commits. @manilabui - this changes the popover's render gate and the read-only card; I've kept it to those hunks. Say if it collides with what you have open.

Verified: 4 canMoveEvent and 2 canAttemptMove cases (theirs refused; ours allowed; ours on read-only / cancelled / a preview refused; a guest's drag allowed to begin), 4 popover cases (theirs not editable; ours editable; read-only never; new event always), 5 data-source cases for isMine (no organizer; somebody else's address; the group-calendar id on a calendar the server says is ours, and not on one it says is shared; ownCalendarIds taking only mine), calendar-guest-move-spec (4: the drag turned into the offer only past the threshold, our own drag left alone, an arrow key, silence on a read-only calendar) and the dialog's two answers - with the popover, drag, context-menu, data-source and calendar-RSVP suites 131 passing. 21 mutations across isMine's terms, both gates, the drag/key interception and the dialog, all red. In a live Google account 68 of 939 events carry a group-calendar ORGANIZER; all sit on a calendar the server reports as somebody else's, so they stay not ours there. Lint, prettier and tsc clean. Extracted from the calendar-scheduling branch (the release-notes-worthy behaviour change there since 2026-08-28: attendees can no longer edit meetings they don't organize).

🤖 Generated with Claude Code

[manilabui]

This works as described on my calendar. I'm fine with guests not moving meetings they don't organize, but right now the drag just does nothing. Can a guest who tries to move one get a dialog saying the event can't be rescheduled, with a "Propose a new time" button on it?

38 of my events have a …@group.calendar.google.com organizer. If Google rewrites ORGANIZER on a secondary calendar as #2924 says, are any of those my own meetings, and can isMine take calendar ownership into account?

recurrenceLoaded isn't protected (removing it from the save condition, defaulting it to true, or deleting the load on mount all leave 145 green), and the _loadEditDefaults comment says the form isn't shown until the load has run, but startEditing shows it straight away. Could the double-click-opens-the-editor change come as its own PR?

A possible follow-up, not for this PR: telling apart guests the organizer has allowed to modify the event, and offering them a way to edit it in Google Calendar. The synced copy doesn't carry that permission, so it would need a sync engine update as well.

🤖 Generated with Claude Code

[brhellman]

Rebased onto #2928's new head as adb324d7d.

1. A guest's drag now begins (canAttemptMove: writable calendar, not cancelled, not a preview) and, once it is past the drag threshold, the grid clears it and shows a dialog - "This meeting can't be rescheduled / Only the organizer can move it" - with Propose a new time and Cancel; the arrow keys get the same dialog. offerCounterInsteadOfMove in calendar-rsvp.ts opens the same picker the context menu uses. Specs: calendar-guest-move-spec.ts (drag past the threshold, drag of our own meeting left alone, arrow key, read-only stays silent) and the dialog's two answers in calendar-rsvp-spec.ts.
2. Group-calendar organizers: isMine is now also true when the ORGANIZER is a ...@group.calendar.google.com id and the event sits on a calendar the server (DAV:owner, Record whether each calendar is the account's own or shared into it #2916) says is ours; the data source reads the calendars for that. In Brian's live DB the 68 such events (of 939) are all on "LINBIT Bereitschaft", which the server reports as someone else's, and none list him as an attendee, so they stay not ours there; on a secondary calendar you own they become editable. Four cases in calendar-data-source-spec.ts.
3. recurrenceLoaded is gone: the save already writes the rule only when the Repeat control differs from what was read, so the flag protected nothing a spec could reach. The comment went with it.
4. Double-click opening the editor is its own PR now, stacked on this one, with the startEditing prop and the on-mount read of the rule (spec'd: the rule is read off the event, and not for a new event).

Mutations run: 19 across isMine's three terms, canAttemptMove/canMoveEvent, _isEditable, the drag and arrow-key interception and the dialog's answer; all red.

[brhellman]

The double-click-opens-the-editor change is now its own PR, #2937, stacked on this one.