The proxy manager caches interfaces marshaled in the payload and returns them on QI without a cross-process call.
However, it doesn't know about interfaces which don't exist.
We can determine this from the payload, since interfaces which don't exist will have a null pointer.
We use this information to avoid querying the proxy in this case.
MozReview-Commit-ID: FnzDetmTiPP
--HG--
extra : rebase_source : 076ce714a69d3883a149487e5f9235ff495eedd0
Now that virtual buffers have to render across processes, we want to eliminate as many cross-process calls as possible.
This includes QueryInterface calls, since buffers query for several interfaces on every node they visit.
To avoid these cross-process QI calls, we include interfaces clients are likely to request in the handler payload.
This way, they get marshaled in the single call used to retrieve the object.
This patch does the following:
1. Passes the interceptor when building the payload.
We need this so we can get interceptors for other interfaces.
2. Splits the payload into two main parts: a static part and a dynamic part.
The (new) static part contains the interface pointers. The dynamic part contains the rest.
This is necessary because the refresh call cannot pass the interceptor, but the interceptor is needed to build the static part.
Also, re-building the static part is pointless when refreshing.
3. Includes the interface pointers in the payload (BuildStaticIA2Data).
The pointers also have to be cleaned up after marshaling.
4. Releases the interface pointers in the handler after the payload is received.
We do this because they're aggregated by the proxy manager as they're unmarshaled.
MozReview-Commit-ID: 6VRLMNScgwW
--HG--
extra : rebase_source : 249589643b7a69e870962ea55a44849bf03a2693
If QueryHandlerInterface returns E_NOINTERFACE, the proxy will be queried for the interface.
However, the handler might know that the interface is definitely not available and could thus avoid a pointless cross-process call.
To facilitate this, the handler can now return S_FALSE to signal that the proxy should not be queried, thus immediately returning E_NOINTERFACE to the client.
MozReview-Commit-ID: 4RtBsA9BTOV
--HG--
extra : rebase_source : 4b0dcb16c469361c1944b24568ceb83fd0ac09c1
The flag of force update is only for fallback items. Fallback items can't share the images with other items.
So I replace 'aForceUpdate' with 'aFallback' and remove some unnecessary checks.
MozReview-Commit-ID: Dcu95FZXlUz
--HG--
extra : rebase_source : 1ec96e0b5d39cc5760b4b5490c5ca9ccbc1933c6
We hit assertion which were added by bug 1414713. That tells us an ancient bug.
There is comment which claims that we should move all list items in the right
list node to left list node if the list nodes should be merged. However, this
has never been done actually since 2002. The code tried to move *some* list
items in the right list node to the list list node. However, retrieving first
list item at an offset almost always failed because the offset variable has
never been initialized.
If we believe the comment, we should move all children of the right list node
to the left list node. However, until we get a testcase to reach this case,
we should keep current behavior, i.e., do nothing, for unexpected regression.
MozReview-Commit-ID: 1r81q1m44oW
--HG--
extra : rebase_source : fc02520f76440d5900fec0517aa6bcd188e1a14f
This brings bitflags 0.9.1 to clap, which removes bitflags 0.7.0.
Source-Repo: https://github.com/servo/servo
Source-Revision: 4d5ff42751ed25ebdc3f298d89acb7eaf58572e5
--HG--
extra : subtree_source : https%3A//hg.mozilla.org/projects/converted-servo-linear
extra : subtree_revision : 36bb86929c90eadca1f49981d472a96bbf2d9c1a
Remove GeckoLayerClient.java since it's no longer used, and update
auto-generated JNI bindings. r=me for trivial patch.
MozReview-Commit-ID: CiNPLLkh3VJ
--HG--
extra : rebase_source : 0cddae4f8914dd6f6a0631a7e69701828476f2b8
Use LayerSession::Compositor or NativePanZoomController to call back
into Java from nsWindow.
MozReview-Commit-ID: 9JUKAE5XQK1
--HG--
extra : rebase_source : 9d6861b9d641e6abb6b6f94f5f9c4b3661ff071c
Move the GeckoLayerClient JNI callbacks to LayerSession.Compositor. Move
the rest of the viewport code to LayerView. Finally, move the input
event synthesis code to NativePanZoomController.
MozReview-Commit-ID: 1FEAM43KcwL
--HG--
extra : rebase_source : af2ded170a79f13afbb1f690ae466e77c7145ff6
Instead of sending an event through the global EventDispatcher in
GeckoLayerClient, switch to using the per-GeckoView EventDispatcher in
GeckoInputConnection, to handle scroll-to-focused-input-on-resize. This
lets us implement the same functionality for standalone GeckoView.
The patch also fixes some small bugs including unregistering
not-registered events, not scrolling when switching input focus, and
inadvertent scrolling when not showing the keyboard.
MozReview-Commit-ID: 20OZP9dMXtI
--HG--
extra : rebase_source : d9dee0fd8b3d01147b8b2eda5154c380d0f167dd
Patch gets rid of `LayerView.getMatrixForLayerRectToViewRect`, and just
uses `LayerView.getZoomFactor` directly when calculating the matrix in
GeckoInputConnection. This also lets us avoid the `isCompositorReady`
call on a non-UI thread. To get the correct offset, we need the screen
bounds from Gecko, so it's passed to Java as the first element in the
rect array. Using bounds from Gecko lets us avoid having to deal with
things like the dynamic toolbar animator ourselves.
MozReview-Commit-ID: 6I61SZGyQyO
--HG--
extra : rebase_source : d576fb7ef9a42de10b14db662e5c4833f16f6312
- This makes all of them return an enum, quite similar to rust.
- This makes all of them derive from a single implementation of an
integer comparison.
- This implements that integer comparison in terms of simple operations,
rather than "smart" computation, which turns out to allow compilers
to do better optimizations.
See https://blogs.msdn.microsoft.com/oldnewthing/20171117-00/?p=97416
--HG--
extra : rebase_source : 89710d7954f445d43e6da55aaf171b1f57dce5fc
This PR implements cancellation for fetch, and uses it for XHR. This means that fetch clients can now send a message to the fetch task asking for the network request to be aborted.
Previously, clients like XHR had abort functionality but would implement it by simply ignoring future messages from the network task; and would not actually cancel the network fetch.
Source-Repo: https://github.com/servo/servo
Source-Revision: 00b3612fe9045f1cb6ba6c89ba7801642a6fa2c9
--HG--
extra : subtree_source : https%3A//hg.mozilla.org/projects/converted-servo-linear
extra : subtree_revision : 4ab32f3b8477c9f2157490eeeb76683e5a5d98f7
<!-- Please describe your changes on the following line: -->
Work in progress, and not quite worth a review yet. (Continuation of https://github.com/servo/servo/pull/4117)
TODO
- [ ] cache metadata (find some subset of`net_traits::Metadata` that can be shared across threads, it seems the problem is mainly stuff inside `hyper::header` in the `headers` field)
- [ ] determine which other fields of a `Response` need to be cached, so a full and valid one can be returned upon a cache hit.
- [ ] determine how to best share the cache across fetch threads (inside HttpState like I tried now?)
- [ ] Spend more time reading the spec and make sure the cache follows it where it matters.
- [ ] Make the current wpt tests pass.
- [ ] More...
---
<!-- Thank you for contributing to Servo! Please replace each `[ ]` by `[X]` when the step is complete, and replace `__` with appropriate data: -->
- [ ] `./mach build -d` does not report any errors
- [ ] `./mach test-tidy` does not report any errors
- [ ] These changes fix#12972 (github issue number if applicable).
<!-- Either: -->
- [ ] There are tests for these changes OR
- [ ] These changes do not require tests because _____
<!-- Also, please make sure that "Allow edits from maintainers" checkbox is checked, so that we can help you if you get stuck somewhere along the way.-->
<!-- Pull requests that do not address these steps are welcome, but they will require additional verification as part of the review process. -->
Source-Repo: https://github.com/servo/servo
Source-Revision: e2bc0f017cfd24734777f64d8607b23dbe9552d7
--HG--
extra : subtree_source : https%3A//hg.mozilla.org/projects/converted-servo-linear
extra : subtree_revision : 059fa41ef6c333ab9cc0adece4f431fec59a84c2
<!-- Please describe your changes on the following line: -->
Initially, we're just running test-perf on linux-nightly, but at some point we should run on other architectures.
---
<!-- Thank you for contributing to Servo! Please replace each `[ ]` by `[X]` when the step is complete, and replace `__` with appropriate data: -->
- [X] `./mach build -d` does not report any errors
- [X] `./mach test-tidy` does not report any errors
- [X] These changes do not require tests because they're test infrastructure
<!-- Also, please make sure that "Allow edits from maintainers" checkbox is checked, so that we can help you if you get stuck somewhere along the way.-->
<!-- Pull requests that do not address these steps are welcome, but they will require additional verification as part of the review process. -->
Source-Repo: https://github.com/servo/servo
Source-Revision: 721723291cfc4bf6ee2facd2b919446038fa0c94
--HG--
extra : subtree_source : https%3A//hg.mozilla.org/projects/converted-servo-linear
extra : subtree_revision : 7cd48d859821bb419920e68e6bc38d754bf4721e
This drops the abuse of make dependencies to factor our steps
in l10n repacks and language packs.
It's replaced with clobbers, libs, and the packaging, which
can be shared between installers and the language pack.
MozReview-Commit-ID: 2vEHp0QogXQ
--HG--
extra : rebase_source : c14f0b17904f8f9e0e3f004384c2a292ceeafe4f
extra : source : e95f8bd7b5190bf1d3fae75a88a15ebbe6613809
For regular builds, we build the mar from an unpackaged exe on windows.
For repacks, we just built that from our l10n-stage directory,
don't unpack again.
MozReview-Commit-ID: 8gQ9G23RgzB
--HG--
extra : rebase_source : b3eb944a0cf423a4b0e22d8884fc90780a3bb109
extra : source : 84091f931dff9e1253b0cfa96b4197255a94ddb2
To reduce any contention from migrating multiple resources at the
same time, it seems prudent to wait for each to finish before
migrating the next. res.migrate is only sometimes properly async,
so the simplest thing to do today is to just always await
completeDeferred.promise. Additionally I am filing Bug 1418455
to track simplifying this to simply await res.migrate.
MozReview-Commit-ID: LdgNsxfuzu4
--HG--
extra : rebase_source : d111110f1445ed84785523ee12b5afb3da2d7207
I'm adding a test for internal properties after this in Gecko.
Source-Repo: https://github.com/servo/servo
Source-Revision: e67bb6ccb5a4fa64f5aaea76cfe76943e873c7c5
--HG--
extra : subtree_source : https%3A//hg.mozilla.org/projects/converted-servo-linear
extra : subtree_revision : 5a281b0b961cb73d8837b11393fd63fc33e3dae7
The "linear time" comment was rendered incorrect by bug 1404181.
And as the body is now only returning a member variable, it's more efficient
and maintainable to just have it to the header.
MozReview-Commit-ID: 4vjB1PyemxR
--HG--
extra : rebase_source : 8089dd66218922a07ffcca9a35f9809522752255
Can now crop screenshots to a given list of XUL elements, which is specified by CSS selectors or custom functions.
Also changed behavior so that if different window types are given, the application exits.
MozReview-Commit-ID: CqmIJFufONw
--HG--
extra : rebase_source : ede33b1c25a8507cdc6abcc6f4bc697e3acd9f0d
