This commit is a prerequisite for part 5 of this patch set
and introduces the `text directive user activation` flag
as well as the necessary boilerplate to transport the value
through process boundaries in `nsDocShellLoadState` and `LoadInfo`.
There is no changed behavior in this commit.
Differential Revision: https://phabricator.services.mozilla.com/D212816
This is a fairly significant patch, however it would be difficult to break it
down into smaller patches:
1) The various mechanisms used to manage ContentParent lifecycles have been
merged together into a common "KeepAlive" system. A process will
begin shutdown when its keepalive count reaches 0. (though it will
still wait for all BrowserParents to also be dead before sending the
actual shutdown message as before).
This replaces a number of bespoke systems for tracking BrowserParent
instances in different lifecycle states, remote workers, ongoing
process switches, and preallocated processes.
2) KeepAlives are now managed automatically by a UniquePtr variant
(Unique[Threadsafe]ContentParentKeepAlive). This makes the hand-off
over KeepAlive lifecycles explicit, even for workers.
3) All KeepAlives are now keyed by a BrowserId, which will be 0 for keepalives
not associated with a specific tab. This allows the new process
selection logic to count all tabs other than the one being navigated
when deciding which process to use.
4) The process switching logic now tracks it's KeepAlive with a BrowserId,
meaning that ongoing process switches are considered when performing
process selection, even if the BrowserParent hasn't been created yet.
Differential Revision: https://phabricator.services.mozilla.com/D213338
This is a fairly significant patch, however it would be difficult to break it
down into smaller patches:
1) The various mechanisms used to manage ContentParent lifecycles have been
merged together into a common "KeepAlive" system. A process will
begin shutdown when its keepalive count reaches 0. (though it will
still wait for all BrowserParents to also be dead before sending the
actual shutdown message as before).
This replaces a number of bespoke systems for tracking BrowserParent
instances in different lifecycle states, remote workers, ongoing
process switches, and preallocated processes.
2) KeepAlives are now managed automatically by a UniquePtr variant
(Unique[Threadsafe]ContentParentKeepAlive). This makes the hand-off
over KeepAlive lifecycles explicit, even for workers.
3) All KeepAlives are now keyed by a BrowserId, which will be 0 for keepalives
not associated with a specific tab. This allows the new process
selection logic to count all tabs other than the one being navigated
when deciding which process to use.
4) The process switching logic now tracks it's KeepAlive with a BrowserId,
meaning that ongoing process switches are considered when performing
process selection, even if the BrowserParent hasn't been created yet.
Differential Revision: https://phabricator.services.mozilla.com/D213338
The code path that was used for loading a document
with fission disabled did not take user activation into account.
This made almost all WPTs for Text Fragments fail,
since they depend on this flag to decide if
a text directive is allowed to be scrolled to.
This patch makes sure that the user activation is
also available for code paths that run through
`CanonicalBrowsingContext::FixupAndLoadURIString()`
to load a document.
This is done by adding a flag `hasValidUserGestureActivation`
to the `nsIOpenWindowInfo`, which carries it through
the JS stack for loading a document.
Differential Revision: https://phabricator.services.mozilla.com/D213401
This commit is a prerequisite for part 5 of this patch set.
The idea is to keep the information whether the last
load done by the doc shell was a same-document navigation.
The value is reset for every navigation, therefore it
should always be correct.
There is no changed behavior in this commit.
Differential Revision: https://phabricator.services.mozilla.com/D212818
This commit is a prerequisite for part 5 of this patch set.
This flag is set in the docshell, based on `user gesture activation`.
This is not necessarily sufficient,
as the text fragments spec bases the value of
this flag also on User Involvement, which is not
implemented in Gecko yet.
The flag is then plumbed through the parent process
(`DocumentLoadListener::CreateDocumentLoadInfo()`)
to the document, where it can be consumed.
This includes client-side redirects as described in [0].
There is no changed behavior in this commit.
[0] https://wicg.github.io/scroll-to-text-fragment/#restricting-the-text-fragment
Differential Revision: https://phabricator.services.mozilla.com/D212817
This commit is a prerequisite for part 5 of this patch set
and introduces the `text directive user activation` flag
as well as the necessary boilerplate to transport the value
through process boundaries in `nsDocShellLoadState` and `LoadInfo`.
There is no changed behavior in this commit.
Differential Revision: https://phabricator.services.mozilla.com/D212816
This is a fairly significant patch, however it would be difficult to break it
down into smaller patches:
1) The various mechanisms used to manage ContentParent lifecycles have been
merged together into a common "KeepAlive" system. A process will
begin shutdown when its keepalive count reaches 0. (though it will
still wait for all BrowserParents to also be dead before sending the
actual shutdown message as before).
This replaces a number of bespoke systems for tracking BrowserParent
instances in different lifecycle states, remote workers, ongoing
process switches, and preallocated processes.
2) KeepAlives are now managed automatically by a UniquePtr variant
(Unique[Threadsafe]ContentParentKeepAlive). This makes the hand-off
over KeepAlive lifecycles explicit, even for workers.
3) All KeepAlives are now keyed by a BrowserId, which will be 0 for keepalives
not associated with a specific tab. This allows the new process
selection logic to count all tabs other than the one being navigated
when deciding which process to use.
4) The process switching logic now tracks it's KeepAlive with a BrowserId,
meaning that ongoing process switches are considered when performing
process selection, even if the BrowserParent hasn't been created yet.
Differential Revision: https://phabricator.services.mozilla.com/D213338
This patch makes some adjustments to follow the spec more strictly by
separating highlighting a text fragment (ie. adding the ranges to a `eTargetText` Selection)
from scrolling to it.
Scrolling to the text fragment now follows the steps given in [0] more closely,
in particular now the closest common ancestor of the target range is being used
as `:target` element.
This change also sets the focus to the start of the first text fragment, as indicated by the spec.
This resets the normal selection on the page.
The wpt test in /css/css-pseudo/target-text-005.html is based on the idea of having a same-doc
text fragment navigation with a selection.
This test fails with this patch applied, because the selection is reset.
It is currently unclear what behavior is correct here, thus the test is kept for now.
[0]: https://wicg.github.io/scroll-to-text-fragment/#invoking-text-directives
Differential Revision: https://phabricator.services.mozilla.com/D211025
This patch makes some adjustments to follow the spec more strictly by
separating highlighting a text fragment (ie. adding the ranges to a `eTargetText` Selection)
from scrolling to it.
Scrolling to the text fragment now follows the steps given in [0] more closely,
in particular now the closest common ancestor of the target range is being used
as `:target` element.
This change also sets the focus to the start of the first text fragment, as indicated by the spec.
This resets the normal selection on the page.
The wpt test in /css/css-pseudo/target-text-005.html is based on the idea of having a same-doc
text fragment navigation with a selection.
This test fails with this patch applied, because the selection is reset.
It is currently unclear what behavior is correct here, thus the test is kept for now.
[0]: https://wicg.github.io/scroll-to-text-fragment/#invoking-text-directives
Differential Revision: https://phabricator.services.mozilla.com/D211025
This changes where the IsUpgradeDowngradeEndlessLoop check triggers.
Before this patch, it triggered during the redirect caused by the https
upgrade. With this patch, it triggers during the downgrade for http
redirects. META and JS redirect are still detected during upgrade.
This should be fixed as a follow up (See Bug 1896691).
Downgrade in this context means same url, except with the scheme http
instead of https.
Different query parameters normally lead to different responses by web servers.
Don't consider the '#ref' part of the uri, because it doesn't get send to
the server and therefore can't change the server response.
We can't use the redirect chain anymore, because the query parameters
are trimmed since Bug 1715785.
This also removes the config option dom.security.https_only_check_path_upgrade_downgrade_endless_loop,
because it adds unnecessary complexity. Removing it for this patch is
easier.
https-only, https-first and httpssvc_https_upgrade tests had to be
modified, because they depended on the incorrect handling of query
strings in loop detection.
Differential Revision: https://phabricator.services.mozilla.com/D193672
This changes where the IsUpgradeDowngradeEndlessLoop check triggers.
Before this patch, it triggered during the redirect caused by the https
upgrade. With this patch, it triggers during the downgrade for http
redirects. META and JS redirect are still detected during upgrade.
This should be fixed as a follow up (See Bug 1896691).
Downgrade in this context means same url, except with the scheme http
instead of https.
Different query parameters normally lead to different responses by web servers.
Don't consider the '#ref' part of the uri, because it doesn't get send to
the server and therefore can't change the server response.
We can't use the redirect chain anymore, because the query parameters
are trimmed since Bug 1715785.
This also removes the config option dom.security.https_only_check_path_upgrade_downgrade_endless_loop,
because it adds unnecessary complexity. Removing it for this patch is
easier.
https-only, https-first and httpssvc_https_upgrade tests had to be
modified, because they depended on the incorrect handling of query
strings in loop detection.
Differential Revision: https://phabricator.services.mozilla.com/D193672
This patch refactors how we check for text formats when deciding how to handle
resources, such that more text MIME types will be rendered in-browser, rather
than downloaded.
This change requires us to move more away from using the Gecko-Content-Viewers
category in the category manager for this decision, as we need to handle an
unlimited number of MIME types behind the scenes.
Support for Gecko-Content-Viewers was left in for both the in-tree use for
application/http-index-format and dynamically determining whether image/avif
and image/jxl are supported, as well as for the message/rfc822 type used by
Thunderbird.
Differential Revision: https://phabricator.services.mozilla.com/D212078
Thunderbird architecture does not currently allow for redirecting channels to
child processes, so new schemes need to be isolated to the parent process.
`x-moz-ews` was selected to clearly designate that the scheme is private and
restricted to Mozilla/MZLA code. It doesn't follow the RFC 7595 recommendation
to use a registered domain as a namespace, but this was felt to be unwieldy.
Differential Revision: https://phabricator.services.mozilla.com/D212439
This patch refactors how we check for text formats when deciding how to handle
resources, such that more text MIME types will be rendered in-browser, rather
than downloaded.
This change requires us to move more away from using the Gecko-Content-Viewers
category in the category manager for this decision, as we need to handle an
unlimited number of MIME types behind the scenes.
Support for Gecko-Content-Viewers was left in for both the in-tree use for
application/http-index-format and dynamically determining whether image/avif
and image/jxl are supported, as well as for the message/rfc822 type used by
Thunderbird.
Differential Revision: https://phabricator.services.mozilla.com/D212078
`PresShell::GetRootScrollFrameAsScrollable()` is equivalent to
`PresShell::GetRootScrollContainerFrame()`.
In ScrollContainerFrame.h, `DecideScrollableLayer()` has two versions, one has
four parameters, and the other has five parameters with the fifth parameter
`aDirtyRectHasBeenOverriden` having a default value `nullptr`. When we switch
the caller from `nsIScrollableFrame` to `ScrollContainerFrame`, we need to
remove the default value for the fifth parameter to avoid ambiguity.
Differential Revision: https://phabricator.services.mozilla.com/D211494
In theory, changing the return type from `nsIFrame*` to `ScrollContainerFrame*`
exposes `ScrollContainerFrame` to the callers who might not needed, but almost
all of the callers in cpp files are already exposed to `nsIScrollableFrame`, as
demonstrated in this patch via replacing the #include from
"nsIScrollableFrame.h" to "ScrollContainerFrame.h", so this is OK.
Some callers can be simplified since we no longer need `do_QueryFrame` to
`nsIScrollableFrame`.
Differential Revision: https://phabricator.services.mozilla.com/D211488
`PresShell::GetRootScrollFrameAsScrollable()` is equivalent to
`PresShell::GetRootScrollContainerFrame()`.
In ScrollContainerFrame.h, `DecideScrollableLayer()` has two versions, one has
four parameters, and the other has five parameters with the fifth parameter
`aDirtyRectHasBeenOverriden` having a default value `nullptr`. When we switch
the caller from `nsIScrollableFrame` to `ScrollContainerFrame`, we need to
remove the default value for the fifth parameter to avoid ambiguity.
Differential Revision: https://phabricator.services.mozilla.com/D211494
In theory, changing the return type from `nsIFrame*` to `ScrollContainerFrame*`
exposes `ScrollContainerFrame` to the callers who might not needed, but almost
all of the callers in cpp files are already exposed to `nsIScrollableFrame`, as
demonstrated in this patch via replacing the #include from
"nsIScrollableFrame.h" to "ScrollContainerFrame.h", so this is OK.
Some callers can be simplified since we no longer need `do_QueryFrame` to
`nsIScrollableFrame`.
Differential Revision: https://phabricator.services.mozilla.com/D211488
This patch addresses the problem that we currently collect HTTPS-First telemetry
for sites that are not reachable at all, be it through always causing a error or
through always timing out.
- On a downgrade, do not collect telemetry instantly, but instead save the
telemetry data in the load state for the downgraded request
- That telemetry data will then be copied over into the document load listener
of the new request
- On a successful request, if we have downgrade data in the load listener, we
collect the downgrade telemetry, as the downgrade seems to have been
successful
- Similar to the downgrade case, we only count the upgrade metric once we
encounter a successful request annotated with the information that it was
upgraded by HTTPS-First, instead of counting it instantly on the decision to
upgrade. This also means the upgrade metric will not include loads that are
downgraded again anymore
- Add a testcase for a site which is neither reachable via HTTP nor HTTPS, and
ensure no telemetry is collected
Differential Revision: https://phabricator.services.mozilla.com/D210792
Same-document navigation follows a different code path than normal navigation
and was therefore not covered in the initial implementation for text fragments.
Same-document navigation does not set a URI in the `Document`, which
is the way cross-document navigation would parse text directives from the URL.
Instead, `nsDocShell::ScrollToAnchor()` is called via
`nsDocShell::InternalLoad()`-> `nsDocShell::HandleSameDocumentNavigation()`.
This code path needs to parse and remove the fragment directive from the new
fragment to be able to find text fragments and to allow for element-id fallback.
`nsDocShell::ScrollToAnchor()` needs to start an attempt to scroll to the text fragment
if it exists. It must not, however, clear the uninvoked text directives, because a
same-document navigation could happen before the document is fully loaded,
hence the target text might not be part of the DOM tree.
As per spec, a second attempt to scroll to the text fragment is done after the load
is completed. This is done by `Document::ScrollToRef()`, which is called by
`nsDocumentViewer::LoadComplete()` after the load has finished.
This call will clear the uninvoked directives.
Differential Revision: https://phabricator.services.mozilla.com/D209726
This change utilizes components.conf to distinguish between the contract implementations of nsISessionStoreFunctions per platform.
Differential Revision: https://phabricator.services.mozilla.com/D206904
These properties were previously being checked much later, during xptcodegen,
and causing methods to be treated as noscript implicitly.
This change enforces the noscript requirements earlier when possible in
xpidl.py, to produce better errors, and adds addiitonal checks to ensure that
types which need to be builtinclass are marked as such.
This required some changes to interfaces in order to satisfy the new checks.
Differential Revision: https://phabricator.services.mozilla.com/D207804
Previously the `boolean` type was also declared using a `bool` typedef in
xpidl, meaning that both were used in various places. This patch standardizes
on the built-in `boolean` type, removing the typedef.
Differential Revision: https://phabricator.services.mozilla.com/D206382
- Create a local page we will access via an about: URL
- In that page, demonstrate how to do some stuff, then fire a custom event
indicating its stuff is done _and_ return complex data in that event.
console.log() didn't seem to be visible in the HiddenWindow, so I also
piped out debug strings for development purposes, so they can be
console.logged in the Service, where we can see them.
- Instead of listening for DOMContentLoaded/pageshow, instead listen for
the new CustomEvent.
- In UserCharacteristicsPageService, receive the data from the page and
populate the appropriate Glean metrics
- Refactor the calling of nsUserCharacteristics::PopulateData() and
SubmitPing(). Now we call PopulateDataAndEventuallySubmit() which will
eventually call SubmitPing after our promise is resolved.
- To make it a little cleaner (so ContentPageStuff() isn't calling
SubmitPing()) we return the promise out of ContentPageStuff() that
PopulateDataAndEventuallySubmit() will await and then call SubmitPing()
when that promise resolves
Differential Revision: https://phabricator.services.mozilla.com/D203055
After bug 306344 this is the last remaining incompatibility with
<marquee>. I think we should be able to drop it. Use counters for
"finish" are relatively high (>0.1%), but I'm pretty sure that's because
that event name is shared with web animations. "bounce" is a more
realistic measure, and that's 0.003%, which is more expected.
Differential Revision: https://phabricator.services.mozilla.com/D206387
In particular:
* Always handle meta viewport in RDM. This fixes bug 1625999 too by
making touch simulation enabled and disabled consistent.
* Restore the resolution to 1 when toggling RDM. This is just simpler,
and we're not keeping around the visual viewport offsets anyways
so...
* Deal with the change more easily, at the same point we switch
scrollbars etc.
Differential Revision: https://phabricator.services.mozilla.com/D206266
We also simplify the tooltip text provider loop, and make it work better
now that e.g. the anonymous button in file inputs can be targeted.
Differential Revision: https://phabricator.services.mozilla.com/D203759
We don't want to start skipping entries just because callers passed in
aRequireUserInteraction, since the feature is not ready and the flag in the
entries might not be correct.
Differential Revision: https://phabricator.services.mozilla.com/D203376
If session history in the parent is enabled then session store only works
correctly if the platform collection code is turned on.
Differential Revision: https://phabricator.services.mozilla.com/D203375
These prefs don't do anything unless browser.sessionstore.platform_collection is
enabled, but we don't need the granularity that they provide. Let's just use
have the one browser.sessionstore.platform_collection pref control everything.
Differential Revision: https://phabricator.services.mozilla.com/D203374
This patch modifies the behaviour of loads when the DONT_RETARGET
nsIURILoader flag is set, making them ignore the Content-Disposition
header. This means that loads which cannot trigger downloads will
attempt to display handleable content which would otherwise be
downloaded.
This keeps overall behaviour of object/embed elements more similar to
their behaviour pre-Fission, while allowing them to load attachment PDFs
and Images as-if they were being displayed by a plugin.
This patch does not change the existing behaviour around
unknown/unhandleable resource types in object/embed elements.
In Gecko, object/embed elements are prevented from triggering downloads
or external protocol handlers during their initial load. Other browser
engines can trigger a download for an unknown resource type (or
sometimes an attachment resource).
The new pref dom.navigation.object_embed.allow_retargeting can be
enabled to instead trigger a download when loading these resources
within an object/embed element.
Differential Revision: https://phabricator.services.mozilla.com/D201645
We also simplify the tooltip text provider loop, and make it work better
now that e.g. the anonymous button in file inputs can be targeted.
Differential Revision: https://phabricator.services.mozilla.com/D203759
This changes comes with several different refactorings all rolled into one,
unfotunately I couldn't find a way to pull them apart:
- First of all annotations now can either recorded (that is, we copy the value
and have the crash reporting code own the copy) or registered. Several
annotations are changed to use this functionality so that we don't need to
update them as their value change.
- The code in the exception handler is modified to read the annotations from
the mozannotation_client crate. This has the unfortunate side-effect that
we need three different bits of code to serialize them: one for annotations
read from a child process, one for reading annotations from the main process
outside of the exception handler and one for reading annotations from the
main process within the exception handler. As we move to fully
out-of-process crash reporting the last two methods will go away.
- The mozannotation_client crate now doesn't record annotation types anymore.
I realized as I was working on this that storing types at runtime has two
issues: the first one is that buggy code might change the type of an
annotation (that is record it under two different types at two different
moments), the second issue is that types might become corrupt during a
crash, so better enforce them at annotation-writing time. The end result is
that the mozannotation_* crates now only store byte buffers, track the
format the data is stored in (null-terminated string, fixed size buffer,
etc...) but not the type of data each annotation is supposed to contain.
- Which brings us to the next change: concrete types for annotations are now
enforced when they're written out. If an annotation doesn't match the
expected type it's skipped. Storing an annotation with the wrong type will
also trigger an assertion in debug builds.
Differential Revision: https://phabricator.services.mozilla.com/D195248
This patch modifies the behaviour of loads when the DONT_RETARGET
nsIURILoader flag is set, making them ignore the Content-Disposition
header. This means that loads which cannot trigger downloads will
attempt to display handleable content which would otherwise be
downloaded.
This keeps overall behaviour of object/embed elements more similar to
their behaviour pre-Fission, while allowing them to load attachment PDFs
and Images as-if they were being displayed by a plugin.
This patch does not change the existing behaviour around
unknown/unhandleable resource types in object/embed elements.
In Gecko, object/embed elements are prevented from triggering downloads
or external protocol handlers during their initial load. Other browser
engines can trigger a download for an unknown resource type (or
sometimes an attachment resource).
The new pref dom.navigation.object_embed.allow_retargeting can be
enabled to instead trigger a download when loading these resources
within an object/embed element.
Differential Revision: https://phabricator.services.mozilla.com/D201645
Most are brought over straightforwardly, their Telemetry callsites reworded
to use Glean, with mirroring to the Telemetry probes taken care of by the Glean
Interface For Firefox Telemetry (see the telemetry_mirror property).
There were two special cases: PERF_DOM_CONTENT_LOADED_FROM_RESPONSESTART_MS,
and PERF_DOM_CONTENT_LOADED_MS. Both were removed as they were unused.
Differential Revision: https://phabricator.services.mozilla.com/D200867
In order to use the modifiers in nsGlobalWindowOuter::OpenInternal, add accessor
methods to BrowsingContext, Document, and WindowContext.
Those accessors behave in the same way as ConsumeTransientUserGestureActivation
and RevisePopupAbuseLevel, except for checking the PopupBlocker state.
Differential Revision: https://phabricator.services.mozilla.com/D197862
Associate modifier keys to user activation, in order to use the modifiers
in `window.open` initiated by the user activation.
Depends on D197859
Differential Revision: https://phabricator.services.mozilla.com/D197860
This is somewhat complex, untested, never shipped, and predates
fission.
Remove it to simplify the code.
Depends on D198183
Differential Revision: https://phabricator.services.mozilla.com/D198184
