We weren't handling the error exception object in the Error previewer for
some error classes, and as a result we'd show an undefined grip in the console,
instead of a proper error message.
A test is added to check that we do print wasm errors as expected.
Differential Revision: https://phabricator.services.mozilla.com/D98469
Those traits are going to be removed in Bug 1680280,
and not through the regular backward-compatibility cleanup routine.
Differential Revision: https://phabricator.services.mozilla.com/D98483
This turns all existing backward compatibility comments (or
comments mentioning a specific Firefox version), into the new
syntax we agreed upon in Bug 1673535.
Differential Revision: https://phabricator.services.mozilla.com/D97395
Before showing the pause overlay, we wait for the page to be fully loaded.
This introduced some delay, and at the moment we were showing the overlay, the
debugger could not be paused anymore.
This patch adds another check to ensure the debugger is paused, as well as a
test to avoid regression."
Differential Revision: https://phabricator.services.mozilla.com/D97778
In Bug 1673328, the signature of the resume function was modified,
but we missed that the function was called from the paused overlay,
which means we didn't pass the correct signature anymore, and buttons
of the paused overlay weren't working at all.
This patch fixes the callsite and adds a mochitest where we go
through a simple case of pausing and then stepping and resuming
by clicking on the overlay buttons.
Since the pause overlay is different from the other highlighters
for which we already had test helpers, we needed to add dedicated
test helpers for the pause overlay, in the test actor.
Differential Revision: https://phabricator.services.mozilla.com/D97757
In Bug 1575068, the message of CORSMissingAllowHeaderFromPreflight and CORSPreflightDidNotSucceed
changed, and the category field was updated in the nsIErrorMessage. As a result, the keys in
errordocs.js needed to be changed, which was rightfully done, but the value was changed as well,
which means the resulting url of the Learn more links were modified as well. I'm not sure if
something changed since then, but those 2 urls now lead to 404 pages on MDN.
This patch fixes those URLs to ensure we do link to the right MDN pages.
Differential Revision: https://phabricator.services.mozilla.com/D97129
This avoids instancing twice the different storage type actors (legacy and resources). In order to keep current server tests working, a pref to force instancing legacy actors has been introduced.
Differential Revision: https://phabricator.services.mozilla.com/D97287
Depends on D97153
DevTools window actors rely on sendQuery for 2 APIs: instantiateTarget and addWatcherDataEntry.
If an exception occurs in the child actor, it will be propagated back to the parent, which will reject the sendQuery.
This can make the code revolving around those APIs very fragile. Since this code is involved in setting up the initial targets, it could be nice to catch errors there to avoid too many potential blank toolbox situations.
Differential Revision: https://phabricator.services.mozilla.com/D97155
DevToolsFrameChild actor excludes about:blank windows to avoid creating unnecessary targets.
In the parent process we are supposed to also exclude about:blank windows when fetching the browsing-contexts that should be debugged.
However the parent-process logic relies on checking the documentURI to exclude about:blank. The actor relies on a docShell flag called hasLoadedNonBlankURI.
The printpreview browser element uses an about:blank window, but still displays a document different from about:blank.
Consequently the parent process check considers this document as valid, because of its documentURI.
But the child actor throws because of the hasLoadedNonBlankURI flag.
This changeset adds an additional check for `window.document.documentURI` in the child actor.
Also adds a new mochitest to check that we can open the BrowserToolbox and inspect the print preview
Differential Revision: https://phabricator.services.mozilla.com/D97153
This method only is async in order to allow callers to wait for a process switch
triggered by the call to `loadURI` to be finished before resolving. With
DocumentChannel, we should never trigger a process switch eagerly like this
again, so we don't need any of the async behaviour here anymore.
This part is largely mechanical changes to tests, removing the `await` calls on
`loadURI`, and a follow-up part will remove the actual async logic from
`BrowserTestUtils.loadURI`.
Differential Revision: https://phabricator.services.mozilla.com/D94641
Change the logic in the devtools/server/actors/highlighters.js file to
use absolute paths instead of constructing them from an internal helper
function. Also modify the calls to `register` at the bottom of the file
according to this change.
Differential Revision: https://phabricator.services.mozilla.com/D96301
This change removes docshell's `mTouchEventsOverride` and replaces it
with a new `BrowsingContext` field `TouchEventsOverrideInternal`.
All uses of the old field have been replaced and an override should
now work under fission when there are cross-origin descendent frames.
Differential Revision: https://phabricator.services.mozilla.com/D96414
We were trying to resume the worker debugger in case we
couldn't connect to the worker. Since connectToWorker can
reject when the worker is closed, we might try to resume a
closed worker, which could cause a crash.
In this patch, we simply check if the worker debugger isn't closed
before resuming it.
Differential Revision: https://phabricator.services.mozilla.com/D95990
The function wasn't used, except in a test that this patch removes.
Since it was also the only use of the getSourcesManager hook, we're able to
remove it as well.
Differential Revision: https://phabricator.services.mozilla.com/D95989
For some reason we return an inlineTextChild when the iframe is remote, so only with fission turned on.
When the iframe isn't remote, WalkerFront.inlineTextChild returns undefined.
Differential Revision: https://phabricator.services.mozilla.com/D95868
Before this commit, the `MOZ_HEADLESS` environment variable was set by
the `--headless` process and inherited by the launched Browser Toolbox
process, hiding the debugger UI. This commit allows to launch the JS
debugger when running under `--headless` by explicitly removing the
`MOZ_HEADLESS` variable from the environment. To do this, a small
tweak to `Subprocess.jsm` is used that allows `null` to signal
"removal", distinct from `undefined` for "ignored".
While here I prefixed the output from the launched Browser Toolbox
process, which helps visually distinguish warnings, etc; and I also
corrected a typo that I stumbled across.
Differential Revision: https://phabricator.services.mozilla.com/D95897
actorhasMethod had several technical limitations making it hard to use consistently. We now removed all the call sites for this method. This changeset removes the method and all its dependencies.
Differential Revision: https://phabricator.services.mozilla.com/D95861
The `dom.worker.console.dispatch_events_to_main_thread` pref is used by platform
code to check if console API messages in the worker thread should be dispatched
to the main thread. If so, the message parameters are cloned, or stringified if
they can't be. This is currently the default behavior.
The pref is checked on the server side and added as a trait to the root actor.
On the client, if the pref isn't true, then we accept messages coming from
worker targets in the console. We can't accept them without condition, otherwise
we would get duplicated message (from the main thread AND the worker thread).
The browser_webconsole_console_logging_workers_api.js test is repurposed for
worker logging since it was disabled on e10s anyway. We add a few test case
to check we can get cached and live message, and that non-clonable object, like
worker scope, are displayed like regular objects when the pref is false.
Differential Revision: https://phabricator.services.mozilla.com/D85397
The SELECTABLE_TEXT state is currently under-utilized and probably wrong, since it simply relies on testing if a hypertext has any text. A more correct reflection of the state should be taken from layout's IsSelectable method.
Note, even if an element is styled `user-select: none`, the text will still be keyboard selectable if it is editable (eg. input or contenteditable), so we should consider that SELECTABLE_TEXT.
Differential Revision: https://phabricator.services.mozilla.com/D95783
Depends on D95685
Keep the isFissionContentToolboxEnabled method but always return true to effectively enable all fission related behavior for all content toolboxes.
Differential Revision: https://phabricator.services.mozilla.com/D95686
This method only is async in order to allow callers to wait for a process switch
triggered by the call to `loadURI` to be finished before resolving. With
DocumentChannel, we should never trigger a process switch eagerly like this
again, so we don't need any of the async behaviour here anymore.
This part is largely mechanical changes to tests, removing the `await` calls on
`loadURI`, and a follow-up part will remove the actual async logic from
`BrowserTestUtils.loadURI`.
Differential Revision: https://phabricator.services.mozilla.com/D94641
Depends on D95299
The actors/stylesheets.js file contained three actors.
This change moves each actor in its own file.
Differential Revision: https://phabricator.services.mozilla.com/D95328
Depends on D95297
After moving actors to individual files, we should also do the same with specs.
Both actors are using shared types, which are moved to a shared types file.
Differential Revision: https://phabricator.services.mozilla.com/D95298
The actors/styles.js file contained two actors.
This change moves each actor in its own file.
A shared actors/utils/style-utils file is also created to host shared helpers.
Differential Revision: https://phabricator.services.mozilla.com/D95297
This was hacking the internals of RDP/protocol.js in order to provide
different names to the actor prototype.
I had to rename existing `skipBreakpoints` getter in order to avoid
name collision with the RDP method of the same name.
I also had to tweak a few method arguments as they weren't following protocol.js spec.
They were called without protocol.js automatic marshalling of values.
Finally, hack something in protocol.js in order to avoid duplicated response for ThreadActor.attach.
Differential Revision: https://phabricator.services.mozilla.com/D94958
We no longer need to manuall call threadActor.exit from Target Actors
as the ThreadActor is managed by the Target Actors, so that managed children
will be destroyed when the target actor is destroyed.
Differential Revision: https://phabricator.services.mozilla.com/D94957
Similarly to the frontend, I'm introducing a mixin in order to share code between at target actors.
We might be able to share more, like sources. But at least we share new addition related to watcher and resources.
This would probably have been easier if Actor were using ES Classes.
Differential Revision: https://phabricator.services.mozilla.com/D94948
Traits support was added to the storage actor in FF80.
Firefox release is now FF81 so we can cleanup the related backward compatibility code.
Differential Revision: https://phabricator.services.mozilla.com/D93081
We have to manually call `ThreadActor#exit` as calling `destroy` won't completely
destroy it, since it does not call `Actor#destroy`, whereas `ThreadActor#exit` does.
Differential Revision: https://phabricator.services.mozilla.com/D94835
Allow-list all Python code in tree for use with the black linter, and re-format all code in-tree accordingly.
To produce this patch I did all of the following:
1. Make changes to tools/lint/black.yml to remove include: stanza and update list of source extensions.
2. Run ./mach lint --linter black --fix
3. Make some ad-hoc manual updates to python/mozbuild/mozbuild/test/configure/test_configure.py -- it has some hard-coded line numbers that the reformat breaks.
4. Make some ad-hoc manual updates to `testing/marionette/client/setup.py`, `testing/marionette/harness/setup.py`, and `testing/firefox-ui/harness/setup.py`, which have hard-coded regexes that break after the reformat.
5. Add a set of exclusions to black.yml. These will be deleted in a follow-up bug (1672023).
# ignore-this-changeset
Differential Revision: https://phabricator.services.mozilla.com/D94045
This patch adds support for dedicated worker targets in the Watcher actor.
Shared and Service workers are not handled yet.
In a similar manner to what we already have for frame targets, we add a worker-helper
file that will communicate with a JsWindowActor pair spawned on each document,
that will manage workers (DevToolsWorkerParent/DevToolsWorkerChild).
For a given document, the DevToolsWorkerChild will enumerate the existing workers
related to it, as well as add an event listener to be notified when workers are
being registered and unregistered, and communicate that back to the DevToolsWorkerParent
on the main thread, so worker targets creation and destruction are notified by
the Watcher actor (via target-available-form and target-destroyed-form events).
When a worker is created, the DevToolsWorkerChild for the document the worker
was spawned from will create a WorkerTargetActor, that will live in the worker
thread (using worker-connector.js), passing it resources the Watcher is currently
listening for. It will also handle communication between the main thread and the
worker thread, when the watcher listen to new resources (or stop watching resources).
A WorkerTargetFront is created so the client can be notified about available
resources (via the resource-available-form event, emitted from the worker target).
Tests are added in the next patches of this queue.
Differential Revision: https://phabricator.services.mozilla.com/D85399
This may prevent logging a few exception or error messages.
But thanks to the other patch of this bug, we should try to stop listening
for sources as soon as the thread actor is destroyed.
Differential Revision: https://phabricator.services.mozilla.com/D94484
We have to manually call `exit` as calling `destroy` won't completely destroy it.
ThreadActor.destroy do not call protocoljs.Actor.destroy, ThreadActor.exit does it.
Differential Revision: https://phabricator.services.mozilla.com/D94483
Allow-list all Python code in tree for use with the black linter, and re-format all code in-tree accordingly.
To produce this patch I did all of the following:
1. Make changes to tools/lint/black.yml to remove include: stanza and update list of source extensions.
2. Run ./mach lint --linter black --fix
3. Make some ad-hoc manual updates to python/mozbuild/mozbuild/test/configure/test_configure.py -- it has some hard-coded line numbers that the reformat breaks.
4. Make some ad-hoc manual updates to `testing/marionette/client/setup.py`, `testing/marionette/harness/setup.py`, and `testing/firefox-ui/harness/setup.py`, which have hard-coded regexes that break after the reformat.
5. Add a set of exclusions to black.yml. These will be deleted in a follow-up bug (1672023).
# ignore-this-changeset
Differential Revision: https://phabricator.services.mozilla.com/D94045
Depends on D92220
Rephrasing of code comments and cleanup of documentation to reflect that `CustomHighlighterActor` is now the only actor which deals with highlighting stuff after removing `HighlighterActor` in D92220. No functionality is changed in this patch.
Differential Revision: https://phabricator.services.mozilla.com/D92221
This patch completely removes the `HighlighterActor` and its accessors.
The rest of the patches in this commit series update the consumers and tests to use the new highlighter manager added to `HighlightersOverlay` in D81526. [Bug 1646028](https://bugzilla.mozilla.org/show_bug.cgi?id=1646028) updated all consumers within the **Inspector** panel to use this approach. Now, we're updating the rest of the consumers outside the Inspector.
### Why remove HighlighterActor?
The changes introduced in D81526 for [Bug 1646028](https://bugzilla.mozilla.org/show_bug.cgi?id=1646028) make it possible to highlight nodes using the `CustomHighligtherActor` with the **Box Model Highlighter** implementation.
The `HighlighterActor` was itself a wrapper around **Box Model Highlighter** and Simple Outline Highlighter, and exposed node picking functionality. Simple Outline Highlighter was removed with [Bug 1650094](https://bugzilla.mozilla.org/show_bug.cgi?id=1650094). The node picking functionality was decoupled from `HighlighterActor`in [Bug 1607756](https://bugzilla.mozilla.org/show_bug.cgi?id=1607756).
After those changes, `HighlighterActor` is no longer special. It can be removed in favor of a single generic actor, `CustomHighligtherActor`, with the Box Model Highlighter implementation.
Differential Revision: https://phabricator.services.mozilla.com/D92220
It looks like we are destroying the target actor immediately, without having time
to call _watchDocshells. And given that removeObserver throws if there is no matching listener
already registered, it prevent detaching the target correctly, and, introduces a leak.
Differential Revision: https://phabricator.services.mozilla.com/D94452
Before, targets were created on-demand, from the frontend, by calling ProcessDescriptor.getTarget
when the root actor informed the frontend of a creation of a content process via processListChanged event
and the notification of a new ProcessDescriptor.
This previous behavior was making so that the target actor was created very late and did not allow
recording for resources (console messages for ex), nor setting breakpoints very early during the process startup.
With this new implementation, the watcher actor allows knowing if the frontend cares about content process target or not.
We should only create these targets if the browser toolbox is opened, and typically not if we are only debugging a tab via a regular web toolbox.
We do that via WatcherActor.watchTarget and sharedData object.
content-process-script.js is loaded in all processes. Process script are loaded quite early during process startup.
Soonish after `sharedData` is passed to the content process.
So that it is hopefully executed before most privileged javascript runs.
Javascript code where we possibly want to debug/observe/set a breakpoint on.
This setup allows us to create the Content Process Target actor much earlier than the previous setup.
When created, the target actor immediately start listening for resources (console message and others) via its `addWatcherDataEntry` method.
In a near future, breakpoints will also be set via this method.
We should also followup to use something else than Process Script and sharedData
in order to create the target earlier, before any JS runs.
Differential Revision: https://phabricator.services.mozilla.com/D65529
For now, we were emitting the "tabDetached" event on message-manager-close (=process shutdown),
or when the DevToolsServerConnection closes (=client closes the connection).
But not when the target is being destroyed by the frontend when calling ResourceWatcher.unwatchTargets,
which is calling process-helper's destroyTargets.
This was introducing test failure (browser_target_list_watchTargets.js) with never resolving RDP requests.
Differential Revision: https://phabricator.services.mozilla.com/D91559
Allow-list all Python code in tree for use with the black linter, and re-format all code in-tree accordingly.
To produce this patch I did all of the following:
1. Make changes to tools/lint/black.yml to remove include: stanza and update list of source extensions.
2. Run ./mach lint --linter black --fix
3. Make some ad-hoc manual updates to python/mozbuild/mozbuild/test/configure/test_configure.py -- it has some hard-coded line numbers that the reformat breaks.
4. Add a set of exclusions to black.yml. These will be deleted in a follow-up bug (1672023).
# ignore-this-changeset
Differential Revision: https://phabricator.services.mozilla.com/D94045
The regression was caused because it incorrectly started treating user
stylesheets as "system" sheets, which can't be modified.
Fix the regression by properly using SharedCssLogic.isAgentStylesheet
instead.
Also fix the check for whether chrome rules apply to handle user
stylesheets and inline style attributes in chrome documents correctly,
see the comment.
Differential Revision: https://phabricator.services.mozilla.com/D93088
Depends on D93817
There should no longer be new message managers during the lifecycle of the actor.
We can simplify the storage actor thanks to this.
Differential Revision: https://phabricator.services.mozilla.com/D93819
Depends on D93817
There should no longer be new message managers during the lifecycle of the actor.
We can simplify the storage actor thanks to this.
Differential Revision: https://phabricator.services.mozilla.com/D93819
Depends on D92220
Rephrasing of code comments and cleanup of documentation to reflect that `CustomHighlighterActor` is now the only actor which deals with highlighting stuff after removing `HighlighterActor` in D92220. No functionality is changed in this patch.
Differential Revision: https://phabricator.services.mozilla.com/D92221
This patch completely removes the `HighlighterActor` and its accessors.
The rest of the patches in this commit series update the consumers and tests to use the new highlighter manager added to `HighlightersOverlay` in D81526. [Bug 1646028](https://bugzilla.mozilla.org/show_bug.cgi?id=1646028) updated all consumers within the **Inspector** panel to use this approach. Now, we're updating the rest of the consumers outside the Inspector.
### Why remove HighlighterActor?
The changes introduced in D81526 for [Bug 1646028](https://bugzilla.mozilla.org/show_bug.cgi?id=1646028) make it possible to highlight nodes using the `CustomHighligtherActor` with the **Box Model Highlighter** implementation.
The `HighlighterActor` was itself a wrapper around **Box Model Highlighter** and Simple Outline Highlighter, and exposed node picking functionality. Simple Outline Highlighter was removed with [Bug 1650094](https://bugzilla.mozilla.org/show_bug.cgi?id=1650094). The node picking functionality was decoupled from `HighlighterActor`in [Bug 1607756](https://bugzilla.mozilla.org/show_bug.cgi?id=1607756).
After those changes, `HighlighterActor` is no longer special. It can be removed in favor of a single generic actor, `CustomHighligtherActor`, with the Box Model Highlighter implementation.
Differential Revision: https://phabricator.services.mozilla.com/D92220
This patch adds support for dedicated worker targets in the Watcher actor.
Shared and Service workers are not handled yet.
In a similar manner to what we already have for frame targets, we add a worker-helper
file that will communicate with a JsWindowActor pair spawned on each document,
that will manage workers (DevToolsWorkerParent/DevToolsWorkerChild).
For a given document, the DevToolsWorkerChild will enumerate the existing workers
related to it, as well as add an event listener to be notified when workers are
being registered and unregistered, and communicate that back to the DevToolsWorkerParent
on the main thread, so worker targets creation and destruction are notified by
the Watcher actor (via target-available-form and target-destroyed-form events).
When a worker is created, the DevToolsWorkerChild for the document the worker
was spawned from will create a WorkerTargetActor, that will live in the worker
thread (using worker-connector.js), passing it resources the Watcher is currently
listening for. It will also handle communication between the main thread and the
worker thread, when the watcher listen to new resources (or stop watching resources).
A WorkerTargetFront is created so the client can be notified about available
resources (via the resource-available-form event, emitted from the worker target).
Tests are added in the next patches of this queue.
Differential Revision: https://phabricator.services.mozilla.com/D85399
Depends on D92756
Detach errors can happen when a target is being destroyed.
Historically, those errors were swallowed in the TargetFront mixin.
However this logic became outdated when we started purging requests during Front::destroy.
This results in non-actionable error messages logged frequently while using DevTools.
This changeset updates the logic in target-mixin.js in order to swallow errors due to purged requests.
Differential Revision: https://phabricator.services.mozilla.com/D92757
Depends on D93030
All Front requests are purged before the Front is destroyed.
In theory there is no reason to process packets after the Front has been destroyed.
It will only create cryptic error messages.
Differential Revision: https://phabricator.services.mozilla.com/D92756
The WebConsoleActor tries to destroy it via a debug:destroy-network-monitor.
But as it is done during the destruction of the target actor, typically when there is a
top level target switching, this message doesn't reach the parent process.
So, ensure destroying this actor from the parent process side, when we see that the
frame target is being destroyed.
Differential Revision: https://phabricator.services.mozilla.com/D92654
- Rename `id` to a more explicit `forwardingPrefix` property
- Add comments to why we need to handle a list of connections
- Remove the `rpcs` property from the connections object
- Turn `rpcDeferred` from an array to an object, as it's used like one.
Differential Revision: https://phabricator.services.mozilla.com/D92752
The previous code was all wrong and we weren't correctly listening to resource
for already existing additional targets. Top level target was still working fine
thanks to WatcherActor.watchResources calling watchTargetResource directly.
Differential Revision: https://phabricator.services.mozilla.com/D91969
The problem with `getOverflowCausingElement`s seems to be that it returned an array
of nodes that weren't properly attached to their parents.
To mitigate this, we call `attachElements` on the entire list of nodes and
return a `disconnectedNodeArray` which ensures that the returned nodes
are properly attached and therefore has all their parent node information
when calling `showNode`.
Differential Revision: https://phabricator.services.mozilla.com/D92360
This patch adds a simple WorkerTargetActor, which is similar to the mock target
actor we had in startup/worker.js.
Its spec file is empty at the moment, and it does not have an associated front,
but that might change with Bug 1633712 where we will at least have events for
resources available/destroyed.
Differential Revision: https://phabricator.services.mozilla.com/D91128
This renames the current WorkerTargetActor which is actually a descriptor.
We also rename the existing `workerTargetFront` occurences to `workerDescriptorFront`
to avoid confusion when we do introduce the real worker fronts and actors.
Differential Revision: https://phabricator.services.mozilla.com/D91100
This patch adds a simple WorkerTargetActor, which is similar to the mock target
actor we had in startup/worker.js.
Its spec file is empty at the moment, and it does not have an associated front,
but that might change with Bug 1633712 where we will at least have events for
resources available/destroyed.
Differential Revision: https://phabricator.services.mozilla.com/D91128
This renames the current WorkerTargetActor which is actually a descriptor.
We also rename the existing `workerTargetFront` occurences to `workerDescriptorFront`
to avoid confusion when we do introduce the real worker fronts and actors.
Differential Revision: https://phabricator.services.mozilla.com/D91100
This renames the internal -moz-math-script-level property in order to
prepare for full math-depth support. Currently, the property is guarded
under a disabled-by-default flag, so there should be no observable
behavior change.
Differential Revision: https://phabricator.services.mozilla.com/D91285
This renames the internal -moz-math-script-level property in order to
prepare for full math-depth support. Currently, the property is guarded
under a disabled-by-default flag, so there should be no observable
behavior change.
Differential Revision: https://phabricator.services.mozilla.com/D91285
This renames the internal -moz-math-script-level property in order to
prepare for full math-depth support. Currently, the property is guarded
under a disabled-by-default flag, so there should be no observable
behavior change.
Differential Revision: https://phabricator.services.mozilla.com/D91285
Nicolas Chevobbe
Julian Descottes
Hubert Boma Manilla
Oriol Brufau
Fabien CASTERS
Zac Svoboda
Micah Tigley
Belén Albeza
Alexandre Poirot
Emilio Cobos Álvarez
Noemi Erli
Nika Layzell
Karl Tomlinson
Cody Welsh
Steven MacLeod
Andrey Bienkowski
Nick Alexander
Neil Deakin
Eitan Isaacson
Yura Zenevich
Razvan Maries
Csoregi Natalia
Daisuke Akatsuka
Cosmin Sabou
Ricky Stewart
Bogdan Tara
Ben Delevingne
Razvan Caliman
Dorel Luca
JulianWels
Mihai Alexandru Michis
Sylvestre Ledru
Gabriel Luong
Gijs Kruitbosch
Frederic Wang
Narcis Beleuzu