Update imports to use FOLDER/index, instead of relying on nodeJS
require behavior (which will look for the index file when importing FOLDER)
Differential Revision: https://phabricator.services.mozilla.com/D198423
Also remove devtools/client/debugger/src/utils/editor/source-editor.js which wasn't
used an look like a relic of ancient times.
Differential Revision: https://phabricator.services.mozilla.com/D198409
I'm introducing a new dedicated method on SourceMapLoader so that we can:
* ensure retrieving the resolved source map URL, even in case of errors
so that we can easily open the source map file from the frontend.
* pass all useful data in one worker call (ignore urls + sources + error + resolve source map url)
Display the source map error in a new dedicated warning footer in the bottom of the editor pane.
(Removing failing jest test. Mochitests should cover the source maps.)
Differential Revision: https://phabricator.services.mozilla.com/D188586
Extract information out of the source map worker to be stored into the Debugger store.
Renamed SourceMapLoader getOriginalURLs to loadSourceMap to better translate the heavy process involved in this method.
While doing that, also inline the transfer of source map ignore list in this call (instead of always having another worker message right after).
This will be used and tested in the next changeset.
Differential Revision: https://phabricator.services.mozilla.com/D188585
The main goal here is to re-throw the exception so that the debugger can start displaying source map errors.
Note that the previous changeset tweaked the StyleEditor code to handle exceptions.
The Debugger will be handled in the next changeset.
Differential Revision: https://phabricator.services.mozilla.com/D187576
* Accept exceptions from `getOriginalURLS` (see next changeset).
* Avoid instantiating an editor for minified stylesheet when we have a functional original stylesheet.
We were doing that solely for getting their `friendlyName`, whereas we could lazily compute it only when we need it.
Differential Revision: https://phabricator.services.mozilla.com/D187575
I haven't been able to repro this so far. The failure screenshot shows
the tooltip but not the toolbox, which is quite odd, too... Hopefully
this helps figure out what's going on.
Differential Revision: https://phabricator.services.mozilla.com/D198540
This fixes the issue allowing the result list to scroll as the user presses the up/down to move through
the list items. Seperating this patch from the test (as the test is quite tricky), so we can land the fix.
Differential Revision: https://phabricator.services.mozilla.com/D193721
We shouldn't show the popup for keywords (`const`, `let`, `for`, …).
codeMirror marks those with the `cm-keyword` class, so we can ignore those in `_isInvalidTarget`,
but we need to exclude `this`, for which we do want to show the preview.
Differential Revision: https://phabricator.services.mozilla.com/D197585
I'm introducing a new dedicated method on SourceMapLoader so that we can:
* ensure retrieving the resolved source map URL, even in case of errors
so that we can easily open the source map file from the frontend.
* pass all useful data in one worker call (ignore urls + sources + error + resolve source map url)
Display the source map error in a new dedicated warning footer in the bottom of the editor pane.
(Removing failing jest test. Mochitests should cover the source maps.)
Differential Revision: https://phabricator.services.mozilla.com/D188586
Extract information out of the source map worker to be stored into the Debugger store.
Renamed SourceMapLoader getOriginalURLs to loadSourceMap to better translate the heavy process involved in this method.
While doing that, also inline the transfer of source map ignore list in this call (instead of always having another worker message right after).
This will be used and tested in the next changeset.
Differential Revision: https://phabricator.services.mozilla.com/D188585
The main goal here is to re-throw the exception so that the debugger can start displaying source map errors.
Note that the previous changeset tweaked the StyleEditor code to handle exceptions.
The Debugger will be handled in the next changeset.
Differential Revision: https://phabricator.services.mozilla.com/D187576
* Accept exceptions from `getOriginalURLS` (see next changeset).
* Avoid instantiating an editor for minified stylesheet when we have a functional original stylesheet.
We were doing that solely for getting their `friendlyName`, whereas we could lazily compute it only when we need it.
Differential Revision: https://phabricator.services.mozilla.com/D187575
browser_html_tooltip_variable-height.js has a 300px tall tooltip, along
with 600px tall content, which overflows.
The click outside of the tooltip goes to the content unless it's
clipped... I'm not sure that's the intention of the test but it seems to
match the pre-existing behavior.
MANUAL PUSH: Trivial orange fix CLOSED TREE.
This adds proper borders and radii on Windows and Linux, which are
otherwise not present, and simplifies a bit the code by using the same
code path across platforms, and between HTML and XUL.
Remove some ::-moz-focus-inner styles which no longer apply to un-themed
buttons.
The extra -0.5 offset I think was compensating for the margins and
missing border-box sizing.
Differential Revision: https://phabricator.services.mozilla.com/D197793
Unmounting the top level React component will help avoid trigerring React updates
when async thunk actions keep dispatching new actions after the tool is closed.
This was leading to various exceptions in many React components.
Also avoid a race condition where `responsiveFront` could be undefined when closing
RDM while it is still initializing.
Finally, ensure registering the `ResponsiveUI` immediately in `activeTabs` to prevent
race condition in openIfNeeded.
Adding some test coverage for such race condition, but can't enable this on verify.
In chaos mode, there is too many exceptions at the test teardown.
Ideally, RDM should:
* only close the RDP client and let the server restore the default value on its own.
i.e. we shouldn't do any request on closing.
* RDM close requests should wait for previous initialization request to be done.
This would avoid zillion of possible exceptions because initialization is very async
and each step may throw because of things are already being destroyed.
But this wouldn't address race condition when the tab is closed.
So RDM should probably try to have a more resilient initialization sequence as well.
Differential Revision: https://phabricator.services.mozilla.com/D197581
A link points to a blog post explaining why we made the change.
Once the notice is closed by the user, it won't be opened again.
The notice is not localized so it's easier to uplift for now; it will
be localized in a follow up.
Depends on D197897
Differential Revision: https://phabricator.services.mozilla.com/D197924
This patch updates the `NotificationMessage` element in `notificationbox.js` so that it extends our newer `moz-message-bar` component instead of the deprecated `message-bar` component. Many of the changes are just dealing with the implications of making things async (so that we can ensure `moz-message-bar.mjs` gets imported). I tried to break out places where I modified related code and tests into separate patches to mitigate some of the review pain here.
This patch solves a longstanding issue where we were loading `in-content/common-shared.css` in the chrome since it gets used by the `message-bar` element. It also makes some small visual changes to our infobars (slight outline, icon colors, adds a bit of spacing).
Differential Revision: https://phabricator.services.mozilla.com/D189872
These two settings are only available via the web console commands as they expect number as arguments
which is hard to implement via the context menu on the debugger button.
Move stdout logging to a dedicated method as onEnterFrame reached eslint complexity limit.
Differential Revision: https://phabricator.services.mozilla.com/D196832
The JavaScript Tracer may be initiated by the debugger, or stoped by itself when reaching some limit.
In these cases, the start and stop messages wouldn't have been logged in the console.
Differential Revision: https://phabricator.services.mozilla.com/D196831