Added 'onKeyDown' event handler with a 'stopPropagation()' method preventing the 'keydown' event from triggering on the parent element.
Differential Revision: https://phabricator.services.mozilla.com/D60718
--HG--
extra : moz-landing-system : lando
When we set a thread to directory root, `item.path`, which is equal to `thread.actor` in a thread's case, is used as the project directory root and stored in `prefs`. When debugger is closed and reopened, `thread.actor` will be different. But the debugger state is initialized with an old directory root. For example, directory root is `server0.conn0.child1/thread19` from the last time, but thread is now "server0.conn1.child1/thread19."
I think it's error-prone to bump up `connX` by 1 when we see a `thread.actor` as `prefs.projectDirectoryRoot`. This patch removes the storage for threadActors, meaning when we set a thread to root, we don't keep track of it in `prefs`. Then we reopen the debugger, it doesn't show a thread as root. For other directories, `prefs` still works.
{F1862523}
Differential Revision: https://phabricator.services.mozilla.com/D58064
--HG--
extra : moz-landing-system : lando
When uplifting localization changes to beta, language packs might only be updated a few days later.
With the current implementation of DevTools startup, any uplift of a change to key-shortcuts.properties will completely prevent devtools from starting.
Differential Revision: https://phabricator.services.mozilla.com/D52078
--HG--
extra : moz-landing-system : lando
In bug 1551585 we added a new inactivity warning shown on top,
right, bottom, left and z-index properties when they are used on elements
that are not positioned.
Turns out that z-index *can* be used on unpositioned elements if these
elements are grid or flex items.
In grid layouts it's useful to determine which item shows on top when
several overlap in a grid cell.
In flex layouts, z-index values other than auto create a stacking context
even if position is static.
So this change simply avoids showing the warning for z-index if the element
is a flex or a grid item.
Differential Revision: https://phabricator.services.mozilla.com/D60673
--HG--
extra : moz-landing-system : lando
This patch moves the pref setting of "devtools.responsive.browserUI.enabled"
within the add_task functions, so that the prefs are in a consistent state
throughout the task.
Differential Revision: https://phabricator.services.mozilla.com/D60590
--HG--
extra : moz-landing-system : lando
This commit changes the testing utility to be generic to work for any
input radio or checkbox on the page, where the input is located near
in the DOM to the text element.
Differential Revision: https://phabricator.services.mozilla.com/D58466
--HG--
extra : moz-landing-system : lando
This commit creates profiler presets for the current browser. It does not
handle remote profiling yet, as there is no UI surface for it yet. The
preset setting updates happen inside of the reducers and presets, and not
the selectors. This design was chosen so that as the presets were changed,
it would actually change the settings below, making it easy to change and
customize the settings by chosing the nearest preset.
The UI designs are still in a bit of a flux, so this UI design was the
easiest to implement for the initial pass, but we don't yet have consensus
on what the UI will look like for the final design.
Differential Revision: https://phabricator.services.mozilla.com/D58465
--HG--
extra : moz-landing-system : lando
We weren't displaying results when they were
falsy (0, false, empty string, ...). This is
fixed by this patch that adds a few test cases
to ensure we don't regress.
Differential Revision: https://phabricator.services.mozilla.com/D60692
--HG--
extra : moz-landing-system : lando
The fix for Bug 1478156 has made the color picker keyboard accessible. In the process, the color swatch element in the Rules view was changed from a `<span>` into a `<button>` element.
There is some strange interaction between markup structure, CSS properties and the default styling for `<button>` which causes an unwanted line break after the button. It happens when doing a multi-line text selection to copy multiple CSS declarations to the clipboard. The bug was raised by many people on different occasions (Bug 1598686, Bug 1600519, Bug 1609891).
After a lot of investigation, I was unable to identify the exact cause for this unwanted line break after color swatches in clipboard content.
The solution I ended up using is reverting from `<button>` back to `<span>` for the swatch element. To preseve the keyboard accessibility, a "keydown" event handler is set to trigger the color picker when pressing SPACE or RETURN when the swatch element is focused. As an added benefit, all other swatches have been made keyboard accessible with the same behavior.
Differential Revision: https://phabricator.services.mozilla.com/D60515
--HG--
extra : moz-landing-system : lando
The react-dom file is only handled when it originates from node_modules.
This means we won't add it to the React group in stacktrace if it's loaded
as an independent file, like it's done in the devtools codebase for instance.
This patch fixes that and adds a few test cases to ensure this works
as expected.
Differential Revision: https://phabricator.services.mozilla.com/D60652
--HG--
extra : moz-landing-system : lando
Summary
After speaking with some of the l10n guys it is clear that whenever we don't want things translated then it shouldn't be in the `.properties` file. With this in mind I have moved the table header strings into `settings.js`.
Sorry @jdescottes, that means asking you for review again.
Differential Revision: https://phabricator.services.mozilla.com/D59837
--HG--
extra : moz-landing-system : lando
Summary
I needed to make changes to `devtools/client/shared/test/browser_tableWidget_mouse_interaction.js` because it was affected by the tiny change in structure to the table. This test is full of things like `table.tbody.firstChild.firstChild.firstChild` so it should be updated at some point but I am stuck for time at the moment.
Differential Revision: https://phabricator.services.mozilla.com/D59496
--HG--
extra : moz-landing-system : lando
This patch make it so that we don't set the
terminalEagerResult to null as soon as we
get the SET_TERMINAL_INPUT action (except
if the expression is falsy).
This means the eager evaluation result element
will only be updated when we get the eager
evaluation results.
Differential Revision: https://phabricator.services.mozilla.com/D60252
--HG--
extra : moz-landing-system : lando
There's no need to make a call to the server when
we know we don't expect anything from it.
Differential Revision: https://phabricator.services.mozilla.com/D60250
--HG--
extra : moz-landing-system : lando
