I wasn't around when RFPTarget::SiteSpecificZoom was implemented, but the current implementation is very confusing.
In Firefox, we have two modes of zoom level. Per tab zoom, and site specific zoom. It is controlled in two (maybe more) places. browser-fullZoom.js and in [CanonicalBrowsingContext.cpp](https://searchfox.org/mozilla-central/rev/ac81a39dfe0663eb40a34c7c36d89e34be29cb20/docshell/base/CanonicalBrowsingContext.cpp#285-289).
Our current implementation disables site specific zoom in `browser-fullZoom.js` by checking the RFP target, but it doesn't modify `CanonicalBrowsingContext.cpp`. So,`CanonicalBrowsingContext` still thinks we are using site specific zoom.
Pre-bug1914149 this method worked because we didn't keep/inherit zoom level across navigations. Post-bug1914149, it no longer works because we keep the zoom level across navigations in `CanonicalBrowsingContext` and let `browser-fullZoom.js` reset to its correct value back.
The issue is caused because `CanonicalBrowsingContext` keeps the previous page's zoom level, but `browser-fullZoom.js` thinks we use tab zoom mode, so it doesn't bother setting the zoom level for the site/page. So, we end up keeping the zoom level.
The solution here I'm suggesting is, doing the opposite of what we are doing in `browser-fullZoom.js`. So, now we should force SiteSpecificZoom with RFP. The reason I'm suggesting this is because within the same site, even acrss tabs, we persist cookies, so fingerprinting isn't much of concern here. We also don't persist zoom levels in private browsing. So, linking normal to PBM isn't a concern either.
So, in summary,
- If you open the same site, in 100 tabs, all of them will get the same zoom level with this patch (just like default normal Firefox)
- If you are in PBM, the zoom level is NOT persisted.
- If you are in normal browsing, the zoom level is persisted, but so are cookies.
Original Revision: https://phabricator.services.mozilla.com/D257497
Differential Revision: https://phabricator.services.mozilla.com/D261948
This patch fixes a number of issues that clear up a few different
intermittents in browser_tab_preview.js.
1. In browser_tab_groups_tab_interactions_telemetry.js, a call to "close
duplicate tabs" would automatically open the confirmation hint panel.
If not manually closed, it seems to persist in a half-way open state
where the `animating` attribute is always `true`. This is now fixed.
2. There is a test somewhere that seems to apply the
`sidebar.verticalTabs` pref and doesn't turn it off. I would like to
figure out which test is causing this, but as a last defence I have
made sure this is always properly set within this test suite.
3. Some tests check that THP is disabled if another panel is open. These
used to work by opening the `appMenu-popup` menu (the hamburger menu
in the top right of the browser). The problem with this approach is
that this panel (and basically every other panel) are very complex
and are prone to change. This was causing an intermittent failure
where attempting to close this panel threw an error in console that
seemed to be specifically related to this panel. I fixed this by
creating a fake minimal panel that is guaranteed to not change as
application code changes. (I borrowed this technique from
https://searchfox.org/mozilla-central/source/browser/base/content/test/popupNotifications/browser_popupNotification_queue.js).
4. There is a suite of tests that checks that THP is disabled in the
event that a drag operation is performed. The problem is that
dragging a tab by 100px sometimes detaches the tab and creates a new
window, breaking the test run. Reducing the drag amount seems to fix
this.
5. As a last line of defence, I have added a state reset function at the
start of each test to ensure that all panels are closed and that the
mouse is not hovering over the tab strip. (The latter was being done at
the end of some tests already, but this makes sure it is consistently
applied everywhere.)
Closes bug1898099, bug1905944, bug1933597.
Differential Revision: https://phabricator.services.mozilla.com/D258976
The required range for the `animDropElementIndex` is between 0 and `this.ariaFocusableItems.length`, inclusive. That maximum value signifies that all tab strip items (and particularly the last one) need to shift out of the way because the user is dragging something to the end of the tab strip.
When dragging past the last tab strip item, `animDropElementIndex == this.ariaFocusableItems.length` and there is no overlapping element. In that scenario, we fall back to setting `dropElement = this.ariaFocusableItems[oldDropElementIndex]` but there is no such element. The `dropElement` is undefined, which ultimately results in calling `Tabbrowser.moveTabsAfter` and appending the dragged tab(s). That happens to be the right result.
However, if you do a short vertical drag, `dropElement` also does not get set, so whatever you were dragging moves to the end of the tab strip. That was bug 1959350 and we tried to fix it by not moving tabs if `dropElement` wasn't set. https://phabricator.services.mozilla.com/D247339
This had the side effect of not making it possible to drop a tab at the end of the tab strip if you were dragging forward. It was still possible to drag forward to the last position, then drag backward slightly to get the front edge of the tab to overlap the last tab in the tab strip, which would allow dropping into the last position.
I think the main thing that would help is to make sure that `dropElement` is always set for valid moves. When dragging past the last tab and there is no overlap, we can fall back to the last non-moving tab strip item.
I needed to swap around how we derive `dropElement` and `newDropElementIndex`. If we fall back to making `dropElement` the last tab, the existing code will try to use `dropElement.elementIndex` as the `newDropElementIndex`, which means the `animDropElementIndex` can never achieve the maximum value that it needs to show all of the tabs moving over. I made it so that we independently determine the `newDropElementIndex` fallback and the `dropElement` fallback without using `newDropElementIndex`.
After this change, `dropElement` should always be a valid non-moving tab strip item if a move should occur. If `dropElement` is undefined, then it's correct that no move should occur.
Differential Revision: https://phabricator.services.mozilla.com/D249715
Fixes a regression that disables tab group management immediately after creating a group with STG off.
I don't quite understand why this comes up, so I hope @ngrato can shed some insight before we land it.
Differential Revision: https://phabricator.services.mozilla.com/D250344
To ensure that a group ends up at the position specified by `index`,
this patch adjusts the index when the tab group moves to the right.
Before this patch, the tab group appeared at a too low (left) index
because the original logic did not account for tabs shifting after a
repositioning to the right.
This also introduces stricter validation for moving tabs near pinned
tabs or other tab groups. Previously, the tabbrowser internals adjusted
the index as needed to fit adjacent to pinned tabs or groups. Now, the
extension API throws an error.
The new behavior matches developer expectations and Chrome's behavior:
https://bugzilla.mozilla.org/show_bug.cgi?id=1963825#c9
Differential Revision: https://phabricator.services.mozilla.com/D249493
The `replaceGroupWithWindow` method currently adopts a tab group in
multiple steps, asynchronously. This can result in externally observable
inconsistencies past initial adoption (see bug).
To fix this, this patch changes the adoption logic by passing the tab
group as the initial item to adopt, and adopts the whole group at once
as needed. Now the logic is similar to drag and drop adoption as
implemented in bug 1908441.
Since tabToAdopt is no longer just a tab, but also a tab group (or even
a tab group label since bug 1908441), the logic in ext-browser.js needs
to be adjusted to make sure that it does not mistake non-tab elements
for tabs. Test coverage is in browser_ext_tabGroups_move_onMoved.js,
as not changing that caused the test to fail with:
> FAIL Tab did indeed move to the new window - {"oldWindowId":3,"oldPosition":"undefined"} deepEqual {"oldWindowId":3,"oldPosition":1} -
Differential Revision: https://phabricator.services.mozilla.com/D248537
To allow extensions to correctly mirror the positions of tabs that were
bulk-moved by moving a tab group, emit the TabMove event in the reverse
order if the tab group moves forwards. This ensures that the previously
recorded index of a tab before the bulk-move remains valid.
Differential Revision: https://phabricator.services.mozilla.com/D248995
In bug 1933744 we prevented SessionStore from reopening a saved tab group into a private window. In bug 1946761 we prevented users from saving tab groups from a private window. The general design principle has been to prevent tab group information leaking between private and non-private windows.
ActionsProviderTabGroups already respected privacy through the implementation of `Tabbrowser.getAllTabGroups`, but `SessionStore.getSavedTabGroups` doesn't respect privacy. Perhaps it should. But in any case, this patch ensures that search results in private windows don't include saved tab groups.
Differential Revision: https://phabricator.services.mozilla.com/D248351
The tabGroups.onRemoved / onCreated / onMoved event depends on the
isAdoptingGroup / removedByAdoption flags to identify tab groups that are
adopted by another window. Although this flag is set for the
`gBrowser.adoptTabGroup` method, it is not in the
`gBrowser.replaceGroupWithWindow` method, which results in incorrect
events.
This patch fixes the issue by setting the isAdoptingGroup flag in
`gBrowser.replaceGroupWithWindow`.
Differential Revision: https://phabricator.services.mozilla.com/D248327
Record when a user ungroups the tabs of a tab group.
We ungroup an entire group when the user selects the "Ungroup tabs" context menu item on the tab group, but we also do that if the user cancels out of tab group creation context menu. We are particularly interested in understanding when users cancel out of the creation process. This may give us a better indicator about accidental tab group creation, e.g. when the user intended to drag and drop to move a tab but ended up creating a tab group from two tabs.
Differential Revision: https://phabricator.services.mozilla.com/D247633
Session restore code doesn't use the standard Tabbrowser.addTabGroup API, so tab groups that enter the DOM via session restore don't raise any TabGroupCreate events. To be consistent, and in order to support addons that need to have an up-to-date view of the tab strip, this patch fires TabGroupCreate on each tab group that enters the DOM.
This patch renames the existing `TabGroupCreate` fired from Tabbrowser.addTabGroup to `TabGroupCreateByUser`. This new event will fire after the tab group enters the DOM (therefore after `TabGroupCreate`) in scenarios where code calls Tabbrowser.addTabGroup with `isUserTriggered = true` to indicate that a user took an explicit action to create this tab group as a new tab group.
Differential Revision: https://phabricator.services.mozilla.com/D246630
This is more visible on Linux because the decorations there are bigger, but I
believe this also fixes issues on windows specially if you have the titlebar
checkbox on.
You need to use mozInnerScreen* coords in order to get the screen coordinates
of the client area.
Differential Revision: https://phabricator.services.mozilla.com/D246829
Tabs can also be unloaded for low memory conditions, so only add the
"discarded" attribute if the tab was unloaded explicitly.
Differential Revision: https://phabricator.services.mozilla.com/D246766
