This reverts commit 524a1fc191.
Revert "Bug 2020707 - Part 5: Gate speculation rules script support behind a pref. r=dom-core-reviewers,jjaschke"
This reverts commit 11741d18ad.
Revert "Bug 2020707 - Part 4: Unregister speculation rules when their script element is removed. r=dom-core-reviewers,hsivonen"
This reverts commit 5202943dc9.
Revert "Bug 2020707 - Part 3: Report an exception if speculation rules parsing fails. r=dom-core-reviewers,jjaschke"
This reverts commit 70c5aa5425.
Revert "Bug 2020707 - Part 2: Add support for speculationrules script type. r=dom-core-reviewers,jjaschke"
This reverts commit 65b30a350b.
Revert "Bug 2020707 - Part 1: Allow speculation rules to be registered to the document from scripts. r=hsivonen,jjaschke"
This reverts commit 28c7050de9.
Revert "Bug 2020704 - Part 2: Add parser for speculation rule sets. r=hsivonen,jjaschke"
This reverts commit eabb8c6eb3.
Revert "Bug 2020704 - Part 1: Add speculation rules crate. r=dom-core-reviewers,jjaschke"
This reverts commit cd65bed005.
Updates:
- Some name changes
- Remove the user-access to SafeResolve via second parameter
(if this returns, it will be under a separate proposal)
Also remove tests that are no longer relevant
Differential Revision: https://phabricator.services.mozilla.com/D305993
This is done by creating a nsID when the object is passed into the content
process. An object containing the nsIChannel instance, as well as information
about where we expect it to be loaded, is kept on the ContentParent actor.
When the actual document is created, this nsID is sent up (wrapped in the
ParentProcessChannelHandle type), and re-matched with the entry in
ContentParent. This allows the actual state to be linked from the
DocumentLoadListener to the final WindowGlobalParent instance.
For now, only the channel is passed up as metadata, with some context as to
where the load is expected to complete being carried along as validation. It
would be good to also validate other information (such as the principal & final
document URI) against this channel object, but that is being left for another
bug.
In the future, we should also likely validate cases where no channel is
provided to ensure it is one of the known cases where a channel is not
available (i.e. initial about:blank, srcdoc, or javascript: URI).
Differential Revision: https://phabricator.services.mozilla.com/D286926
NSS encrypts TLS tokens with an ephemeral key. That key will have rotated when we read TLS tokens
back from storage on a browser restart, causing TLS resumption with those tokens to fail. Disable
TLS token persistence until we can address this in NSS.
Differential Revision: https://phabricator.services.mozilla.com/D306150
This is disabled by default because the HTML parser doesn't support PIs,
but without we miss out on test coverage on unrelated features.
Differential Revision: https://phabricator.services.mozilla.com/D304498
This is useful for nova (which wants to use a gradient border around the
selected tab and right now hack around it with multiple backgrounds).
For now enable only on chrome code by default since we only support
solid borders (which is what we want anyways), and we don't support
combining it with `background-clip: text`.
Differential Revision: https://phabricator.services.mozilla.com/D304960
In anticipation of the next ANGLE update, which will drop support for
shader translator output targetting GLSL versions lower than 150, we
now require a desktop GL context version of 3.2+ to run WebGL. This
means we should avoid setting the REQUIRE_COMPAT_PROFILE flag, as it
only requests a context version of 2.0.
This patch therefore removes the webgl.1.allow-core-profiles pref, and
instead always requests a core context. A consequence of this is that
we will now request core contexts even when the driver does in fact
support compat contexts of a sufficient version, but this should be
harmless.
When a driver does not support desktop GL 3.2, on EGL we will fall
back to attempting to create a GLES 2 context, which may still allow
users to run WebGL 1 on certain hardware.
As a precaution, the WebglAllowCoreProfile gfxVar is retained, which
is used to prevent us creating core contexts on buggy devices. If
this var were to be set false, a compat context would be succesfully
created, but WebGL would be blocked if the supported GLSL version is
insufficient (which would be preferable to creating a core context and
subsequently crashing). In practice, this is currently only set false
on old Mac hardware which is no longer supported.
Differential Revision: https://phabricator.services.mozilla.com/D304818
The `moz_intgemm` pref is enabled by default in Nightly, so this patch doesn't
mark it fuzzing-unsafe as it will be fuzzed anyway (the Python script checks for
this).
Differential Revision: https://phabricator.services.mozilla.com/D303775
With this patch, prefs marked `fuzzing_safe: false` can't be set with `--fuzzing-safe`
in the JS shell. Attempting to set such a pref will print a message to stderr instead.
This seems useful to have for features that are still in development because LLMs are
good at figuring out how to enable things that are still disabled by default.
Differential Revision: https://phabricator.services.mozilla.com/D303774
We replace `layout.css.scroll-driven-animations.viewtimeline.enabled`
with `layout.css.scroll-driven-animations.enabled`.
This preference was added because we used `double` to replace
`CSSNumericValue`. Now we have updated the WebIDL to match the spec, so
we could just use the preference of scroll animations instead. (And so
we enable this interface on Nightly, for the interop purpose.)
Differential Revision: https://phabricator.services.mozilla.com/D304653
We replace `layout.css.scroll-driven-animations.viewtimeline.enabled`
with `layout.css.scroll-driven-animations.enabled`.
This preference was added because we used `double` to replace
`CSSNumericValue`. Now we have updated the WebIDL to match the spec, so
we could just use the preference of scroll animations instead. (And so
we enable this interface on Nightly, for the interop purpose.)
Differential Revision: https://phabricator.services.mozilla.com/D304653
This reverts commit 2f092425d6.
Revert "Bug 2044829 - Only wait for content response on first touch-move when the APZ-aware listener was just discovered via the fast path. r=botond"
This reverts commit 5e1be8b7ba.
Revert "Bug 2041154 - Do not wait the additional content response for touchmove during fast-fling. r=botond"
This reverts commit 0e8e1f4098.
Revert "Bug 2041154 - Factor out the utility function for fast fling. r=botond"
This reverts commit 5aa21e186a.
Revert "Bug 2041154 - Add mochitests for first touch-move content-response wait across the four eApzAwareListeners / eIrregularArea dispatch-to-content cases. r=botond"
This reverts commit b99ea692a9.
Revert "Bug 2041154 - Wait for a content response on the first touch-move if needed. r=botond"
This reverts commit 1007221178.
Revert "Bug 2041154 - Fast-path notify APZ on non-passive APZ-aware listener registration. r=botond"
This reverts commit f7b7a2394b.
Revert "Bug 2041154 - Encapsulate the long-tap touchmove content-response condition in a TouchBlockState method. r=botond"
This reverts commit f1ed0b7d9a.
Revert "Bug 2041154 - Propagate the initial touch move information to InputQueue from APZCTreeManager. r=botond"
This reverts commit 4fadd65737.
Revert "Bug 2041154 - Add mFirstMoveSeen to TouchCounter. r=botond"
This reverts commit e9ef82decb.
kCLAuthorizationStatusNotDetermined indicates an error that MacOS says is
temporary and should be ignored but isn't. We don't know the precise
conditions for this error but it is seen in the wild. When the error is
returned for requests from sites with site permission, we have been presenting
the permissions door hanger anyway, which isn't useful since the user isn't
the reason we were denied. This bug switches to `NoPrompt` (and removes
the related Firefox pref) so that the non-augmented geolocation process runs
for this error. (The user will see the doorhanger unless site permission is
granted, and we won't open system preferences.)
Differential Revision: https://phabricator.services.mozilla.com/D304304
- removes the temporary nightly-only ML and heuristics comparison telemetry
- removes the telemetry preference
- maintains address autofill using ML engine on Nightly only
- switches the ML class to be an instance and not based on AIFeature
- removes the timing and hash computions that are no longer needed
- retains heuristics for credit card detection and ML for address fields, but the autocomplete attribute will override either
- type=email special case applies to both heuristics and ML
- reuses the ML engine each call unless it has been closed in which case a new one is created
- adds a test thta tests the ML specific autofill
Differential Revision: https://phabricator.services.mozilla.com/D302562
velzie