Thank you for contributing to Servo! Please add an `X` inside each `[ ]` when the step is complete, and replace `__` with appropriate data:
- [x] `./mach build` does not report any errors
- [x] `./mach test-tidy` does not report any errors
- [ ] These changes fix #__ (github issue number if applicable).
Either:
- [ ] There are tests for these changes OR
- [x] These changes do not require tests because they change only compilation
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: f8b7e1c6ae66a0198e38fb959cd959a6ef69b84c
This PR fixes#10886. The -p option can be followed by either an interval number or a CSV filename.
* In the interval profiling, the profiler output would be spitted out to the terminal periodically.
Example usage: **./mach run -p 1 http://www.google.com** will print the time-profiling output to the terminal every second.
* In the CSV file profiling, a CSV file will be generate upon termination of servo.
Example usage: **./mach run -x -o out.png -p out.csv http://www.google.com** will generate out.csv upon termination of Servo.
Source-Repo: https://github.com/servo/servo
Source-Revision: d4315855aaddb9d9ed26cd730e9b75e15500c3bd
Addresses issue #10702
the tidy self-test doesn't seem to catch the `member_name:"Foo"` line like it should. the regex follows similar pattern in the file and works in my regex tester tool, so not really sure what is going on there.
Source-Repo: https://github.com/servo/servo
Source-Revision: 8c2c0bd964bfbf1a2d7494bd8fbbdd263173ef7d
This avoids unncessary build script runs caused by changes to unrelated files.
Note: Adds a dependency on https://crates.io/crates/walkdir which is MIT licensed and maintained by BurntSushi.
r? @metajack
Source-Repo: https://github.com/servo/servo
Source-Revision: a85db48623d8f5b5dbddbcc1f1ba0eaa0b531249
* Remove unnecessary dependencies and features from top-level Cargo.tomls. The features for each crate will be computed based on the union of features specified in the dependency graph. Specifying the same ones again just adds more ways for them to get out of sync.
* Move all cargo build environment variables into CommandBase
Fixes#11112. r? @metajack
(Not included: CI test to make sure #11112 doesn't regress again.)
Source-Repo: https://github.com/servo/servo
Source-Revision: 7f76e3ba74a11f1f4bb46f12e17c06270175908f
This preserves the fix from #11097 in the situation where `$CARGO_HOME` and `cargo-home-dir` in `.servobuild` are both present: `.servobuild` should win.
But it changes the behavior when `$CARGO_HOME` is present and `.servobuild` is not: Then `$CARGO_HOME` should be used.
We now check the following values in order of priority and use the first one that is found:
1. `cargo-home-dir` in `.servobuild`
2. `CARGO_HOME` in the environment
3. default value (`<servo-repo>/.cargo`)
r? @larsbergstrom
Source-Repo: https://github.com/servo/servo
Source-Revision: fcebfcc1130b7b3e74f14c8375f825f148e0f152
* Rewrite constructors of `Blob` and `File` with `DataSlice` as argument
* Update WebIDL of `Blob` and `File`
* Implement missing interfaces of `File` (However, due to lack of working `ArrayBuffer/ArrayBufferView` in `Blob`, so it still differs from spec)
* Update WPT test `File-constructor.html`
Source-Repo: https://github.com/servo/servo
Source-Revision: 392135bd0c2f512a0d632a7d76e667bc9af8f4a7
This is a prerequisite for merging #11044, and is an important correctness fix on its own.
r? @Ms2ger
Source-Repo: https://github.com/servo/servo
Source-Revision: 91cabf8427ae8a7693a56c11abfd32b5058c21de
This is intended to support the File API implementation. Basically an event loop with three kinds of messages:
+ Select a file
+ Read a file with ID
+ Delete the ID from manager-owned map
The design decision in this PR is not the final (or best I think) version, welcome reviews :)
TODOs:
- [x] Add multiple file selection
Source-Repo: https://github.com/servo/servo
Source-Revision: b61ad4190f7fb7d4fa32e01a0e3db77a912cb7cf
While here, we make border-*-width round to pixels like Gecko does.
Source-Repo: https://github.com/servo/servo
Source-Revision: 190103a7d7dee0925bb66341d9bc2387a1572a75
<s>Because of the way Gecko stores outline-width in nsStyleOutline::mOutlineWidth (i.e., as its specified value for keywords) we need separate property implementations for servo/gecko products.</s>
The -moz-outline-radius parsing is a bit of a hack; I was just avoiding the effort of factoring out the parsing into a separate method. Let me know if I should do that.
<s>Gecko-side changes are https://bugzilla.mozilla.org/show_bug.cgi?id=1271168.</s>
Source-Repo: https://github.com/servo/servo
Source-Revision: a834bc1ec700ca6447ee13bda9714f78381f2f80
Currently `--profile-trace-path` has no effect if `-p` isn't also passed, because the time profiler doesn't start.
Source-Repo: https://github.com/servo/servo
Source-Revision: 8fab68d6a9543248b1e798dfe53036e33654e8cd
Fixes#4954. r? @jdm
This is based on hyperium/hyper#472, though it doesn't re-use that code directly because Servo configures its own OpenSSL context.
Source-Repo: https://github.com/servo/servo
Source-Revision: 40be84df26ce3ce80851e751374154c015506921
This series of commits fixes#9487, and improves the look of nytimes.com among others.
r? @metajack
Source-Repo: https://github.com/servo/servo
Source-Revision: 1fd9c5583455b873fca1c95b2784f969870073bd
Delete session is needed by the web platform test `webdriver/navigation.py`.
Source-Repo: https://github.com/servo/servo
Source-Revision: 0c673ef3a2adb3ffd8c3181ee4ec5546d610db5b
This use always `cargo-home-dir` in `/.servobuild` even if you set `CARGO_HOME` in your shell.
If you use [racer][racer] with [rustup (multirust.rs)][rustup], you may set `CARGO_HOME` env variable to your shell. Then the previous code would be a problem which does not use `cargo-home-dir` in `/.servobuild`.
[racer]: https://github.com/phildawes/racer
[rustup]: https://github.com/rust-lang-nursery/rustup.rs
Source-Repo: https://github.com/servo/servo
Source-Revision: 0617727f5b58d1c4ebb10cf081f5952b7961fbf4
Part of making the trace-dump.js code unit testable was moving it out to another
repo. This brings in all the changes made while writing unit tests, and making
the code more unit testable.
Unfortunately, there are a lot of whitespace changes; here is the diff ignoring whitespace: https://pastebin.mozilla.org/8870346Fixes#10905.
r? @jdm
Source-Repo: https://github.com/servo/servo
Source-Revision: 835f443865fbaa8784722ad6f950a2c8879fab31
This matches what I believe the OS native defaults to be.
Partially addresses #9487.
r? @metajack
cc @paulrouget
Source-Repo: https://github.com/servo/servo
Source-Revision: 180a9813aa5ecce0a3013cdd30c1ef99ed1d1f6d
I did this to see if it was a significant factor in automated test runs. (Spoiler: it isn't.)
Source-Repo: https://github.com/servo/servo
Source-Revision: 7b8bb0c2485eb46303d97918d91fce73bd98fea9
The default maximum line length Python linters use is 79, so a bunch of editor plugins complain about line lengths, unless you have a tox.ini telling the linter to ignore line lengths (E501) explicitly.
Source-Repo: https://github.com/servo/servo
Source-Revision: c3323f3c70ab53e5fe2f9aae7ba26b1a21cd622a
Gets rid of duplicate versions of bitflags and xml-rs!
Source-Repo: https://github.com/servo/servo
Source-Revision: e94f489e55e0fafed1c88b4d8f833f23545fc48b
LineBreaker calls Fragment::strip_trailing_whitespace_if_necessary and then recalculates the fragment's inline size. But this isn't necessary because strip_trailing_whitespace_if_necessary already recalculates the size.
r? @pcwalton
Source-Repo: https://github.com/servo/servo
Source-Revision: f80641e7a16b832ce755cf936ed9ea6a9289fda9
Five of the BluetoothDevice members has been deleted in the spec:
8d148ba3c3
These were the vendorID, vendorIDSource, deviceClass, productID, productVersion.
Source-Repo: https://github.com/servo/servo
Source-Revision: a153e20c5503b6720d7abbfe36acfdda33a8d4dd
Stops the compiler complaining about `root_pipeline` being an unused variable.
Source-Repo: https://github.com/servo/servo
Source-Revision: 29ed65016047ee10dabf18894ba00ece6227e865
After actually reading a Python tutorial, I realised this can be handled
in a more elegant fashion :-)
Source-Repo: https://github.com/servo/servo
Source-Revision: 283eb4129664593f050e95c092e672a18f4658fc
Currently `compute_layout_damage` does the following for each child of the node it's processing.
1. Update the child with damage from the parent.
2. Update the parent with damage from the child.
When these steps are repeated for the next child, the parent's damage may include flags that came from its previous sibling(s). This means that damage ends up propagating to later siblings, and not just between parents and children as indended.
This patch propagates the same damage to all children, not including any damage from their siblings.
r? @pcwalton
Source-Repo: https://github.com/servo/servo
Source-Revision: bd871d613e9ed1ae4b7f1674275e69ba6da6133b
