In some cases, we were getting null sourceId, which was causing an exception
to be thrown in `getFormattedSourceId` since it expects a string.
We add a simple guard in this function so we don't throw anymore.
But we shouldn't get null sourceId in the first place. It looks like for
some Wasm sources, the SavedFrame sourceId can be 0, which wasn't playing
well with some loose check we were doing in the sources manager.
Finally, the "Copy metadata" button we display when a console message crash
wasn't working. The message now includes the targetFront, which can be serialized
in JSON, so we simply ignore it.
Differential Revision: https://phabricator.services.mozilla.com/D206105
- Guard Services.prefs as this isn't available in workers
- Pass worker's global to getAvailableEventBreakpoints when in worker toolbox
- Check that global exists before checking property on it
Differential Revision: https://phabricator.services.mozilla.com/D207257
`InspectorUtils.getRuleBodyTextOffset` was returning bytes position, and we
were using them directly in Javascript `substring`, which causes problem
with non-ascii chars.
Instead of returning offsets to compute the rule string, we directly return
the string from InspectorUtils which is easier to work with.
Differential Revision: https://phabricator.services.mozilla.com/D204523
When we modify a rule from the inspector, the number of rules and the at-rules
data can't be impacted, so avoid computing this costly information.
Differential Revision: https://phabricator.services.mozilla.com/D205798
Since we're at the end of the cycle, let's fix the regression
by reverting getRuleText to use the JS lexer again.
A test case is added to prevent future regression when we'll
try to move this to an InspectorUtils method again.
Differential Revision: https://phabricator.services.mozilla.com/D204621
We were returning the whole list of rules for a given stylesheet, and then
picking the at-rules we wanted in JS.
This patch make it so that the InspectorUtils method will only return the
at-rules we want directly, so we're building a smaller array of rule in the end.
Since `getStyleSheetRuleCountAndAtRules` also need to return the total number
of rules, this adds a simple `InspectorUtils.getStyleSheetRulesCount` method that
does that.
Differential Revision: https://phabricator.services.mozilla.com/D203878
We were returning the whole list of rules for a given stylesheet, and then
picking the at-rules we wanted in JS.
This patch make it so that the InspectorUtils method will only return the
at-rules we want directly, so we're building a smaller array of rule in the end.
Since `getStyleSheetRuleCountAndAtRules` also need to return the total number
of rules, this adds a simple `InspectorUtils.getStyleSheetRulesCount` method that
does that.
Differential Revision: https://phabricator.services.mozilla.com/D203878
This helps address a performance issue when a bulk or sources are registered.
We could have use findIndexLast, given that the sources are often coming sorted,
but this is not always the case. For example when a new arbitrary source is created by the page.
Also, while the sources look sorted, there is no actual sort being done from the Source Map Worker,
nor by the SOURCE server resource watcher. We would need to add sorting there to be reliable.
In this patch, I'm also avoiding unecessary array in the actor codebase, and, simplifying the load source map actions.
Differential Revision: https://phabricator.services.mozilla.com/D202149
By protocol.js conventions, most target scopes actors have a `targetActor` attribute set to their parent,
i.e. the related target actor. Thread Actor was old enough to not follow this pattern.
Also remove the second constructor argument and `global` attribute as it was unused.
Differential Revision: https://phabricator.services.mozilla.com/D201696
We set up a Map keyed by resourceId, whose value is an AbortController that
controls the MediaQueryList change event listeners that we set on a given stylesheet.
We also take care of aborting the existing abort controller before setting new
event listeners on a stylesheet, to avoid leaks.
Differential Revision: https://phabricator.services.mozilla.com/D200180
With this it's easier to handle existing stylesheets (or to ignore them).
As the stylesheets resource now calls `watch`, we can remove StyleSheetsManager
events that were only used there.
Differential Revision: https://phabricator.services.mozilla.com/D199612
Don't register StyleSheetApplicableStateChanged/StyleSheetRemoved/window-ready events
if they were already registered before.
We take this as an opportunity to control those event listeners
with an AbortController to make it cleaner in destroy.
Differential Revision: https://phabricator.services.mozilla.com/D199611
For now we were releasing object actors one by one.
This would force to send an individual RDP request for each of them.
The console often release all objects actors related to older console message
going over the maximum limit of displayed console messages (10k).
This can easily grow in a large number of actors to be released,
either if console message are receiving many arguments and/or
if many console are logged.
We have to have one request per target as the actors could only be reached
within same-thread actor.
In order to prepare for ObjectFront removal, introduce a target-scoped "Objects" actor
which is a singleton per Target. It will receive the new "release in bulk objects actors"
method. Later, it will start implementing all the existing methods of the Object Actor
in order to migrate away from having to instantiate one Object Front (notice the singular on "Object"),
per inspected JS Object.
On the fronted side a new Object Command is introduced in order to abstract away the RDP/Fronts work.
Differential Revision: https://phabricator.services.mozilla.com/D198784
For now we were releasing object actors one by one.
This would force to send an individual RDP request for each of them.
The console often release all objects actors related to older console message
going over the maximum limit of displayed console messages (10k).
This can easily grow in a large number of actors to be released,
either if console message are receiving many arguments and/or
if many console are logged.
We have to have one request per target as the actors could only be reached
within same-thread actor.
In order to prepare for ObjectFront removal, introduce a target-scoped "Objects" actor
which is a singleton per Target. It will receive the new "release in bulk objects actors"
method. Later, it will start implementing all the existing methods of the Object Actor
in order to migrate away from having to instantiate one Object Front (notice the singular on "Object"),
per inspected JS Object.
On the fronted side a new Object Command is introduced in order to abstract away the RDP/Fronts work.
Differential Revision: https://phabricator.services.mozilla.com/D198784
This revealed that some of the pseudo methods were used outside of the StyleSheetsManager,
so we're making them public instead.
Differential Revision: https://phabricator.services.mozilla.com/D199610
The scrollend event has been content enabled by default for about a
year. Remove the preference that allows the feature to be chrome-only.
Differential Revision: https://phabricator.services.mozilla.com/D197699
There is no need to add all global to the side effect free Debugger.
We should only add the evaluation context globals.
In the typical case it will be the target global.
But for content process targets, it may be multiple globals, which we all need to register.
Also ensure clearing onNativeCall callback to be really sure this is disabled,
even if the Debugger is leaked for some unexpected reason.
Having this listener set may drastically slow down executions by disabling the JIT.
There was no known scenarios where it happened, but it is to be extra safe.
Also simplified a few things in evalWithDebugger as eslint started complaining about its complexity.
Differential Revision: https://phabricator.services.mozilla.com/D195021
This includes the following API:
* Debugger.Object.prototype.createSource
* startColumn field of the parameter object
* Debugger.Script.prototype.startColumn
* return value
* Debugger.Script.prototype.getPossibleBreakpoints
* minColumn and maxColumn property of the query
* columnNumber property of the returned array elements
* Debugger.Script.prototype.getOffsetMetadata
* columnNumber property of the returned array elements
* Debugger.Script.prototype.getOffsetLocation
* columnNumber property of the returned array elements
* Debugger.Script.prototype.getAllColumnOffsets
* columnNumber property of the returned array elements
* Debugger.Script.prototype.getOffsetsCoverage
* columnNumber property of the returned array elements
* Debugger.Source.prototype.startColumn
* return value
This patch modifies DevTools code to convert the column number from/to 1-origin,
while keep using 0-origin on their side.
One exception is the WASM's column number, which had been using 1-origin 1.
Each consumer in DevTools handles the WASM case, and the code can be removed
once DevTools internal also switches to 1-origin column number.
The other exception is to use 1-based column number in logCustomFormatterError,
which is folded from bug 1864783 patch.
Differential Revision: https://phabricator.services.mozilla.com/D193270
This includes the following API:
* Debugger.Object.prototype.createSource
* startColumn field of the parameter object
* Debugger.Script.prototype.startColumn
* return value
* Debugger.Script.prototype.getPossibleBreakpoints
* minColumn and maxColumn property of the query
* columnNumber property of the returned array elements
* Debugger.Script.prototype.getOffsetMetadata
* columnNumber property of the returned array elements
* Debugger.Script.prototype.getOffsetLocation
* columnNumber property of the returned array elements
* Debugger.Script.prototype.getAllColumnOffsets
* columnNumber property of the returned array elements
* Debugger.Script.prototype.getOffsetsCoverage
* columnNumber property of the returned array elements
* Debugger.Source.prototype.startColumn
* return value
This patch modifies DevTools code to convert the column number from/to 1-origin,
while keep using 0-origin on their side.
One exception is the WASM's column number, which had been using 1-origin 1.
Each consumer in DevTools handles the WASM case, and the code can be removed
once DevTools internal also switches to 1-origin column number.
The other exception is to use 1-based column number in logCustomFormatterError,
which is folded from bug 1864783 patch.
Differential Revision: https://phabricator.services.mozilla.com/D193270
Support for breakpoint while unloading the page has been fixed by bug 1806796.
This is only matter of enabling these event types in the UI and covering them by tests.
Differential Revision: https://phabricator.services.mozilla.com/D191979
This will help bug 1569775 and bug 1651522 to update the whole entries of a given data type.
As there is little difference between adding new entries and replacing all entries,
I'm tuning the existing addSessionDataEntry methods to support both behaviors.
Differential Revision: https://phabricator.services.mozilla.com/D192243
These days, after bug 1853323, about:PreferenceStyleSheet is basically
just an empty sheet for most users. For the ones that isn't, it's either
slightly or completely broken (see below).
If we were to implement this today, we'd do it with privileged env()
variables or media queries, and add a bunch of tests for different
combinations, but I don't think it's worth the effort:
* There's no UI for any of these prefs.
* There are no tests for any of these prefs.
* browser.display.focus_ring_or_anything is not great, it's better
replaced by browser.display.show_focus_rings, which actually makes
the :focus-visible pseudo-class work. Otherwise the native controls
may show no rings even though this pref is set.
* for browser.display.focus_ring_style = 0, the color used
(-moz-mac-focusring) is not implemented in any non-macOS platform.
* The ::-moz-focus-inner rules are completely useless nowadays, because
::-moz-focus-inner only works on the windows native theme (which
isn't used on content).
* The focus colors are not color-scheme aware.
So given those, and that if you want the specific current behavior,
these are trivially implementable as a web extension (by using a user
stylesheet), I think we should just remove this code altogether.
Differential Revision: https://phabricator.services.mozilla.com/D190464
Don't check if rules are instances of `CSSGroupingRule` (as `CSSStyleRule` would
match now, so it's not providing the guard we want anymore).
Avoid retrieving stylesheet window and document until it's needed, and only
compute rule line and column for the at-rules we'll return.
Differential Revision: https://phabricator.services.mozilla.com/D185636
Subtle things:
* We now have shadow trees in NAC, inception! Only a couple lines of
code in the style system had to be changed to match :host rules and
so properly.
* Had to make highlighters.css contentaccessible, because otherwise we
can't load it from the shadow tree. I don't think it's a big deal.
* I removed some of the code from highlighters.css that claimed that
stuff inherited from the html element. That's just no longer true.
* Had to switch from `setAttribute("style", ...)` to `.style = ...;`.
This is needed because CSSOM from chrome code bypasses CSP (as
AnonymousContent did), but setAttribute() doesn't, see bug 1424474.
Differential Revision: https://phabricator.services.mozilla.com/D173998