We are still using the nearest neighbor approach as default. This method will be tested along with Kmeans via Nimbus.
From preliminary perf. testing, it's also more performant.
Differential Revision: https://phabricator.services.mozilla.com/D246551
We were initializing two SmartTabGroupingManager instances in tabbrowser and tabgroup-menu. This was
causing issues with telemetry. This is now cleaned up.
Also, adding the model reason for no suggestion per DS request.
Differential Revision: https://phabricator.services.mozilla.com/D246113
Records counts of user interactions with tab groups:
- expand a collapsed tab group
- collapse an expanded tab group
- rename a tab group
- change a tab group's color
- save and close a tab group
- delete a tab group
- reopen a deleted tab group
- ungroup a tab group
- move a tab group to a new window
- reopen a saved and closed tab group from the list all tabs menu
- reopen a saved and closed tab group from a URL bar suggestion
- reopen a recently closed tab group from the recently closed tabs menu
Differential Revision: https://phabricator.services.mozilla.com/D246123
For the first and last tab in a tab group, include the tab group's name in the accessibility description read by screen readers when the tab is in focus. As keyboard users go through the tab strip, the users will receive clues that they are passing into/out of a tab group.
This also reformats the tab tooltips in general.
Before:
```
$title $pid $active - $containerName
$audioPlaying
```
After:
```
$title
$pid $active
$tabGroupName - $containerName
$audioPlaying
```
For each portion of the tab tooltip string, parts of the string might not appear due to configuration (PIDs and active only shown with pref) or user state (container name only shown for a container tab, etc.) or UX choices (don't include the title in the accessibility description because screen readers already read out a tab's title by default)
Differential Revision: https://phabricator.services.mozilla.com/D245186
If a DOM rebuild was scheduled on a previous frame but the rebuild is no longer needed (e.g. due to the "list all tabs" menu no longer being open) then do not clean up nor populate the DOM.
In bug 1953533 we debounced and delayed full DOM rebuilds of the "list all tabs" menu due to severe performance impacts during event-heavy operations like restoring an entire browser session.
When the "list all tabs" menu is closed, it's supposed to get emptied and stop responding to events. However, it was possible to perform actions in the "list all tabs" menu that would trigger events and close the menu at the same time. The triggered event would schedule a "list all tabs" menu rebuild on the next frame, which would then build the "list all tabs" menu DOM despite the menu being closed. When the menu was reopened, the menu builds the menu DOM by inserting new DOM elements, but since the menu was already built, you end up with a doubled version of the tab strip.
Differential Revision: https://phabricator.services.mozilla.com/D246047
This patch adds the most basic tab interaction metrics for tab groups.
As discussed in standup, we agreed to move the `remove_` class of
metrics into its own bug due to extra complexity involved in correctly
capturing these events.
One other thing we discussed was what the scope of events should be for
the `close_` class of events, i.e. should we *only* capture an event
when someone clicks the "X" button or closes from the TOM menu, or
should we also account for things like context menus, keyboard
shortcuts, etc.? Originally we agreed to capture _all_ tab close events
and mark any source that was not one of the above mentioned two as
unknown. However, after thinking about this more, I don't believe this
is the right approach. There are many places in the codebase where a tab
is closed but not because a user deliberately did it (e.g. when moving a
tab to a new window — this is actually done by creating a new tab in a
new window and closing the old). We currently don't distinguish between
user-initiated close actions, so it would take some time to find all
these places and exclude them.
I favour an explicit inclusion approach (which is what we are doing
elsewhere). In this patch I added events for:
- closing a tab from the "X" close button (`close_tabstrip`)
- closing a tab from the tab context menu (`close_tabstrip`)
- closing a tab from the TOM (`close_tabmenu`)
There are other places that could potentially be
addressed, so I suggest moving these to a follow-up bug and addressing
them for 139.
Differential Revision: https://phabricator.services.mozilla.com/D244915
Set aria-setsize on grouped tabs to the number of tabs in the tab group. Set aria-posinset on grouped tabs to the 1-based index of the tab within the tab group. This allows some a11y tools to report to the user that a tab is, for example, tab "2 of 7" in a tab group.
The tab strip uses the `tablist` ARIA role. The ARIA spec and Firefox's a11y engine both forbid nesting `group` ARIA roles inside of `tablist`. As a result, a11y tools always read individual tabs as being tab "X of Y", where Y is the number of tabs in the tab strip and X is the index of the tab in the tab strip. This is good information, but it does not give the user a sense of where a tab is within a tab group.
Differential Revision: https://phabricator.services.mozilla.com/D245185
Tabs outside of tab groups are at the top level (level 1) while tabs inside of tab groups are at the next lower level (level 2). This helps unsighted users get a better sense about the hierarchy of the tab strip.
Differential Revision: https://phabricator.services.mozilla.com/D245184
Tabs outside of tab groups are at the top level (level 1) while tabs inside of tab groups are at the next lower level (level 2). This helps unsighted users get a better sense about the hierarchy of the tab strip.
Differential Revision: https://phabricator.services.mozilla.com/D245184
- record number of saved tab groups
- record min/max/median/average number of tabs in saved tab groups
This patch adds a new observable topic `sessionstore-saved-tab-groups-changed` to SessionStore. This is emitted when a saved tab group is added or removed from the session. BrowserUsageTelemetry subscribes to that topic and updates saved tab group-related metrics.
BrowserUsageTelemetry is also setting those saved tab group-related metrics on startup. BrowserGlue initializes BrowserUsageTelemetry after the session has been restored/set up, so saved tab group information will be available at that time. This is also important to make sure that we keep this metric correct even if a user doesn't touch their saved tab groups during the course of a session.
Differential Revision: https://phabricator.services.mozilla.com/D245138
This patch exposes a groupId field on tabs.Tab objects in extensions.
On mobile (Android), the value is always -1.
On desktop, the value is the numeric representation composed from the
internal ID, which happens to fit in a safe integer. As long as the
browser is not running past year 2255, this works. As a fallback,
when an ID in a different format is encountered, the logic falls back
to returning a new unique ID based on a counter. This ID is then
stored in an internal in-memory map to maintain stable consistency.
Since it is unclear when exactly it is safe to discard this value, this
map is never cleared. This minor memory leak is acceptable because it
does not happen in practice, and even if it does, the number of tab
groups over the lifetime of a Firefox session is not going to reach
excessive values.
Differential Revision: https://phabricator.services.mozilla.com/D245159
When using the tab context menu or drag-dropping tabs to put them into a group, record a metric for the number of tabs added to the group.
Data Science asked us to launch with this metric disabled so that they could control it using server knobs. They expect a high volume of events, so they expect to only enable this metric for some proportion of users.
I converted the existing `TabMove` event derived from `UIEvent` being fired when tabs change their tab index in the tab strip. `UIEvent` doesn't allow for attaching additional context/detail to the event. `TabMove` is now a `CustomEvent` that provides more context about the moved tab and it fires in more cases -- it's possible for the tab index not to change despite the tab having "moved" into/out of a tab group.
This approach would not capture tab movements that occur across multiple frames/event loop iterations.
Differential Revision: https://phabricator.services.mozilla.com/D244616
See details in the bugzilla. The current implementation (removing identical tabs from the suggestions) is confusing to users.
We may uplift this to Beta as well, so the change made is as simple as possible.
Differential Revision: https://phabricator.services.mozilla.com/D244913
Improvements:
1) Preprocess text to remove domain information. This makes similarity check more robust.
2) Optimize for F1, which means increasing threshold and reducing depth parameter
Differential Revision: https://phabricator.services.mozilla.com/D244631
I expect that bug 1955112 was really a duplicate of bug 1954163. Based on that, let's remove this workaround since it's causing trouble.
Differential Revision: https://phabricator.services.mozilla.com/D244564
