This adds support for the `command/resolve` request to support the new Interactive Forms functionality, and updates the refactor processes to use it instead of the original protocol when the client indicates support.
Currently only the "Move to File" refactor uses this functionality (matching the previous version) and it requires the `dart.experimentalInteractiveForms` setting be enabled for it to be used (in case we find bugs while developing the front-end that require incompatible changes).
Fixes https://github.com/dart-lang/sdk/issues/63371
Change-Id: I414a41fa2958ca9dcefe097f35ba28f3fa2fe367
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/508121
Reviewed-by: Samuel Rawlins <srawlins@google.com>
Reviewed-by: Brian Wilkerson <brianwilkerson@google.com>
In a valgrind "profile" of a run of an aot-compiled version of
`pkg/analyzer/tool/stable_analysis.dart` I noticed that
`_OperatorEqualsAndHashCode._hashCode` was called 40,288,281 times,
costing - by itself, i.e. without the actual cost of getting the
underlying hashCode - 723,483,274 instructions - about 1.79% of the
total cost.
Interestingly `_OperatorEqualsAndHashCode._equals` wasn't there, so that
one must be inlined (whereas the `_hashCode` isn't). Stepping via GDB
confirms both things.
Via the inlining tracing in the vm via
```
out/ReleaseX64/dart-sdk/bin/dart \
compile aot-snapshot --verbose \
--extra-gen-snapshot-options="--print_inlining_tree" \
pkg/analyzer/tool/stable_analysis.dart
```
I found
```
Inlining into: 'dart:_compact_hash___Map&_LinkedHashBase&MapMixin&_HashBase&_OperatorEqualsAndHashCode&_LinkedHashMapMixin@3099033_[]='
growth: 0.000000 (9 -> 0)
NO 14 __Map&_LinkedHashBase&MapMixin&_HashBase&_OperatorEqualsAndHashCode@3099033._hashCode@3099033 - Heuristic fail (no small leaf)
NO 16 __Map&_LinkedHashBase&MapMixin&_HashBase&_OperatorEqualsAndHashCode&_LinkedHashMapMixin@3099033._set@3099033 - Not inlinable
```
In `runtime/vm/compiler/backend/inliner.cc` I saw that I could avoid
this "no small leaf" thing by marking it for inlining.
Running stable_analysis through `perf stat` without and with this CL I
get:
```
Without CL:
39,552,232,456 instructions:u
39,552,392,619 instructions:u
39,554,905,123 instructions:u
```
```
With this CL:
39,024,836,004 instructions:u
39,022,158,372 instructions:u
39,022,782,198 instructions:u
```
So this "only" saves something like 527 million instructions (~1.3%),
i.e. less than the ~723 million instructions hoped for, but I'll take
it.
Change-Id: I07a8fb097e46ab48c26bcf7896b12c7312ededc3
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/509000
Reviewed-by: Slava Egorov <vegorov@google.com>
Commit-Queue: Slava Egorov <vegorov@google.com>
- Declare OS::SafeReadMemory in os.h
- Implement using pread64(/proc/self/mem) on Linux and Android
- Add _readNativeMemory VM Service RPC with pre-checks for null
and address overflow
TEST=runtime/vm/service_test.cc
Cq-Include-Trybots: dart/try:vm-asan-linux-release-x64-try,vm-asan-mac-release-arm64-try,vm-asan-win-release-x64-try,vm-dyn-linux-debug-x64-try,vm-dyn-mac-debug-arm64-try,vm-ffi-qemu-linux-release-arm-try,vm-ffi-qemu-linux-release-riscv64-try,vm-linux-debug-arm64-try,vm-linux-debug-ia32-try,vm-linux-debug-simriscv32-try,vm-linux-debug-simriscv64-try,vm-linux-debug-x64-try,vm-linux-debug-x64c-try,vm-linux-release-simarm-try,vm-mac-debug-arm64-try,vm-mac-debug-x64-try,vm-msan-linux-release-x64-try,vm-reload-linux-debug-x64-try,vm-tsan-linux-release-x64-try,vm-tsan-mac-release-arm64-try,vm-ubsan-linux-release-x64-try,vm-ubsan-mac-release-arm64-try,vm-ubsan-win-release-x64-try,vm-win-debug-arm64-try,vm-win-debug-x64-try,vm-win-debug-x64c-try
Change-Id: Id15a82bf478bc4822c08d7fdf0a5c8bfd71a1fe0
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/505781
Auto-Submit: Nourhan Hasan <nourhan.m.hasan@gmail.com>
Reviewed-by: Ben Konyi <bkonyi@google.com>
Reviewed-by: Daco Harkes <dacoharkes@google.com>
Work towards https://github.com/dart-lang/sdk/issues/59524
* avoid_private_typedef_functions - Doesn’t meet any sufficient
criteria? I think a private typedef can certainly make local code
more readable.
* one_member_abstracts - Low impact. This is maybe here for Java
developers? It encourages users to use top-level functions.
Change-Id: I829acda8af0bc2bb265553e1297cd1740761b4da
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/508681
Reviewed-by: Brian Wilkerson <brianwilkerson@google.com>
The resumer in `resume_shutdown_race_test.dart` was running in a tight
`do-while` loop, hammering the VM server with `getVM` and `getIsolate`
requests to check if all isolates are paused at exit. During VM
shutdown, this rapid sequence of requests could cause socket exhaustion,
especially on Windows, or trigger transient connection errors.
This is fixed by adding a 10ms delay using `Future.delayed` at the end
of each iteration, and wrapping the loop's HTTP request block in a
`try-catch` block to gracefully log and ignore transient connection or
request errors during VM shutdown.
Change-Id: I1bfcad7505254a1feb59dfb654421547efc02d46
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/508720
Commit-Queue: Ben Konyi <bkonyi@google.com>
Reviewed-by: Mark Zhou <markzipan@google.com>
Auto-Submit: Ben Konyi <bkonyi@google.com>
Commit-Queue: Mark Zhou <markzipan@google.com>
The CFE constructor tear-off lowerings introduce extra static methods.
Tear-off constants refer to them as `StaticTearOffConstant`s. They
just forward the call to the actual generative, factory or
redirecting factory constructors.
We can avoid these intermediary methods by not using the CFE lowerings.
Change-Id: I1dc1203d6b3023cf1f13bc204a2628ec8a3aebc4
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/508562
Reviewed-by: Nate Biggs <natebiggs@google.com>
Reviewed-by: Srujan Gaddam <srujzs@google.com>
Commit-Queue: Martin Kustermann <kustermann@google.com>
I ran across this bug while testing to see what work was already done.
I didn't take the time to create an issue.
The bug is that an assist was being offered for `class C ^{}` to
convert the class body into a block. The result, for an empty block,
was just to add an extra space before the block body. For a non-empty
block, the result was to delete everything inside the block.
Change-Id: I0342b681ef67b5e659a00126a998d63d593223fd
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/508702
Reviewed-by: Samuel Rawlins <srawlins@google.com>
When `useAuthToken` is true, the VM Service URI includes the
authentication token as a path segment (e.g.,
`ws://127.0.0.1:8181/auth_token/ws`). Previously,
`http_get_isolate_rpc_common.dart` converted `service.wsUri` to an HTTP
`serverUri` by only copying the host and authority, stripping out the
auth token path segments. This caused all subsequent HTTP request
helpers to fail with a missing or invalid authentication code error.
This is resolved by:
1. Enabling `useAuthToken: true` in `http_auth_get_isolate_rpc_test.dart`.
2. Updating
`http_get_isolate_rpc_common.dart` to extract and preserve the path
segments from `service.wsUri` (excluding the trailing `'ws'` segment) and
ensuring a trailing empty segment so that `buildRequestUri` correctly
formats the final HTTP request path.
Change-Id: I773d402af05c451d2ed776a9660c0bc04f4c672c
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/508381
Commit-Queue: Mark Zhou <markzipan@google.com>
Auto-Submit: Ben Konyi <bkonyi@google.com>
Reviewed-by: Mark Zhou <markzipan@google.com>
This change addresses flakiness in SSE tests caused by port conflicts
and race conditions during connection teardown.
Key changes:
- Dynamically allocates ChromeDriver ports in `sse_smoke_test.dart` and `sse_client_test.dart` instead of using a hardcoded port (4444). This avoids conflicts when tests run in parallel on CI bots.
- Adds support for locating ChromeDriver via the `CHROMEDRIVER_PATH` environment variable in both test suites.
- Updates `sse_smoke_driver.dart` to introduce a brief delay before closing connections. This ensures the client-side SseClient has sufficient time to flush its final POST request to the server, avoiding "Bad state: No element" errors.
- Modifies `test_helper.dart` to filter out `--mark_main_isolate_as_system_isolate` from arguments forwarded to spawned testee processes. This is necessary under modern `dart test` runners to ensure processes pause at start as expected.
Change-Id: Idf45cd69f366ec6b6f81cfed955aea337b277dcc
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/508343
Auto-Submit: Ben Konyi <bkonyi@google.com>
Reviewed-by: Mark Zhou <markzipan@google.com>
Commit-Queue: Mark Zhou <markzipan@google.com>
In `http_request_helpers.dart`, `makeHttpServiceRequest` was not closing
the created `HttpClient` instance if an exception was thrown during the
HTTP request, response parsing, or JSON decoding. This could cause
socket leaks in tests that invoke this helper.
This is resolved by wrapping the HTTP request, response, and parsing
logic inside a `try-finally` block, ensuring that `httpClient.close()`
is always executed.
Change-Id: I40d5fb8d2c2cd9c16d83418e71a7dc0d67ef2cdc
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/508740
Auto-Submit: Ben Konyi <bkonyi@google.com>
Reviewed-by: Mark Zhou <markzipan@google.com>
Commit-Queue: Mark Zhou <markzipan@google.com>
Separate the subtype search entry points by result shape and caller
intent. Use `directSubtypeReferences` for reference locations in subtype
clauses, and use `directSubtypesWithMembersOfType` /
`directSubtypesWithMembersOfSubtype` for indexed direct subtypes with
their declared instance members.
Rename `SubtypeResult` to `DirectSubtypeWithMembers` so the model
matches what the search result actually represents. Update the server
member-collection logic to start from direct subtypes of the target type
and then walk direct subtypes by subtype id.
Introduce `SubtypeIndexElementId` as the single representation of
subtype index ids. This keeps id construction consistent between index
building and search, preserves the declaration file for declarations in
parts, and still exposes the element name separately for file-state
prefiltering.
Also make the subtype index expectations text-based so the expected
supertype id, subtype name, and member list are easier to read and
update.
Change-Id: Ice71d9b116aca83372bf595f6e58a8ce4aa9e9d1
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/507780
Reviewed-by: Johnni Winther <johnniwinther@google.com>
Reviewed-by: Samuel Rawlins <srawlins@google.com>
The assist was originall written as a fix, but it's on the list of new
assists that we think users would find useful, so I made it also work
as an assist. The tests are just a copy of the tests for the fix.
The assist is available when the cursor is inside the name of a field
on the left side of a field initializer.
Unlike the lint, which only flags initializers if there is a reference
to a constructor parameter on the right side, the assist will move any
initialization expression. I think that's reasonable, but let me know
if you think we should limit it.
Change-Id: I4f6d90dbd510d5c93f0087933741d298939d4a49
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/508680
Reviewed-by: Samuel Rawlins <srawlins@google.com>
This code seemed a little out of date and non-idiomatic. I used a few modern language features to help it to better comply with our team styles.
* Make declarations private if they can be.
* Make declarations final if they can be.
* Use factory constructors over static methods.
* Do not use type annotations that would be inferred.
* Use patterns for matching JSON data.
* Use extension type when appropriate.
Change-Id: Ib7ecbe51b6d8a94e56a51f84772952362c64049f
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/508424
Commit-Queue: Samuel Rawlins <srawlins@google.com>
Reviewed-by: Brian Wilkerson <brianwilkerson@google.com>
Replace the local marker-based _AstTest helper in ast_test.dart with
ParserDiagnosticsTest parsing and the shared findNode utilities. This
keeps the tests aligned with the inline diagnostics style used by other
parser tests, and avoids a second way to locate AST nodes from marked
source snippets.
Update the affected tests to parse normal source text, then locate the
expected node through findNode single-node accessors. Add a
singleConstructorName accessor so constructor-name tests can use the
same shared lookup path.
Change-Id: Ib8c7c2e7c2c1b320c95810ce148b5a27e68dda43
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/508421
Reviewed-by: Johnni Winther <johnniwinther@google.com>
Commit-Queue: Konstantin Shcheglov <scheglov@google.com>
This adds an internal class hierarch for patterns. This allows for the internal nodes to use InternalVariable rather than Variable and only contain the fields created during body building. This is a step towards separating Variable and InternalVariable fully.
Change-Id: I0e5241bc7163e0771d3420496b1f5ea02b433c87
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/508561
Reviewed-by: Chloe Stefantsova <cstefantsova@google.com>
Commit-Queue: Johnni Winther <johnniwinther@google.com>
Initializers of late local variables are wrapped into closures.
The result type of a closure is a static type of the variable.
So initializer of a late variable with static type Never have a call
to a closure with static result type Never.
TFA summary collector handles calls with static result type Never
specially, treating all subsequent code as unreachable (by setting
control-dependent condition to empty). As a result, all subsequent
code after late variable with static type Never (and initializer) is
tree-shaken.
The fix is to avoid propagating control-dependent condition out of
the late variable initializers.
TEST=pkg/vm/testcases/transformations/type_flow/transformer/regress_63478.dart
Fixes https://github.com/dart-lang/sdk/issues/63478
Change-Id: If41221777c9d249dd21ce5827c586b0f531cbce8
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/508363
Reviewed-by: Slava Egorov <vegorov@google.com>
Commit-Queue: Alexander Markov <alexmarkov@google.com>
The only actual change required to the profiler service is appropriately
retrieving the source position for interpreted frames during processing.
Otherwise, most of the tests just need small changes to account for the
slightly different stack trace generated by interpreted code when
walking the stack as code instead of functions:
* Interpreted code calls the runtime entry DRT_AllocateObject directly
for slow-path allocation instead of via the AllocateObjectSlow stub.
* The current name for bytecode frames when walked as code starts with
"[Bytecode]" instead of "[Unoptimized]".
The test vm/cc/Profiler_GetSourceReport is skipped since:
* It tests compiled code-specific features like classifying source
positions.
* Inserting fake samples for interpreted functions is messier since
many PC offsets may map to the same source position.
The other skipped tests are specific to inlining and/or optimized
compiled code, so there's no point in running them on interpreted code.
TEST=vm/cc/Profiler
Cq-Include-Trybots: luci.dart.try:vm-dyn-linux-debug-x64-try
Change-Id: I3e73f42b2562eb6e3e73fe8d5c3f14389039ee04
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/507082
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Commit-Queue: Tess Strickland <sstrickl@google.com>
The debugger assumes a null suspend state variable in an asynchronous
function means that the function is still in the prologue prior
to setting up the suspend state. However, the interpreter clears the
suspend state variable before returning, and the debugger needs to be
able to pause before returning when single stepping, so earlier a hack
was added to the debugger that detects being at the direct call of the
async return method and/or the return instruction with a null suspend
state variable.
However, there's a much simpler way of ensuring the debugger pauses
before returning: just emit the source position for the return prior to
clearing the suspend state variable. This also ensures that the debugger
still has access to the function's suspend state when pausing before
the return, instead of waiting until it has been cleared and thus is no
longer accessible.
TEST=ci (should not change the result of any current tests)
Cq-Include-Trybots: luci.dart.try:vm-dyn-linux-debug-x64-try
Change-Id: I401cceb169d8692ac379cdc5a531e07cafbe9a65
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/500740
Commit-Queue: Tess Strickland <sstrickl@google.com>
Reviewed-by: Alexander Markov <alexmarkov@google.com>
Given that the standalone target for dart2wasm is feature-complete now,
it makes sense to include it in released SDKs.
This adds the platform and outline files to built SDKs and exposes the
`--standalone` flag in `dart compile wasm`.
It also documents the standalone target in `pkg/dart2wasm/doc`, which
should be helpful as a starting point to use these compiled modules.
Change-Id: I5bd86e9670f03f2955e31789095dd5c462bf149e
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/506920
Reviewed-by: Slava Egorov <vegorov@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Commit-Queue: Martin Kustermann <kustermann@google.com>
Because non-late field initializers in classes with primary constructors have access to the parameters of the primary constructor, field initializers are be moved to the initializer list of this constructor, so the generated AST will no access these parameters out of scope.
This CL add an exception to this rule for mixin classes:
In the non-erroneous cases, these can't have parameters, so the field initializers can stay in the field declaration. This is done to ensure that mixin transformation can
simply clone the mixin class fields, instead of having to fetch the initializer from the initializer list of the constructor.
For mixin classes with parameters in the primary constructor, which is an erroneous case, the initializers are moved to the constructor like for other
primary constructors to avoid generating an AST where the parameters are accessed out of scope.
Change-Id: Iff64992bce88368a524437c1856f50d779bf6a58
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/507423
Commit-Queue: Johnni Winther <johnniwinther@google.com>
Reviewed-by: Konstantin Shcheglov <scheglov@google.com>
This updates the handling of initializers in primary constructor body declarations. The parser is updated to pass the `:` token, similar to what is done for the regular constructors. The CFE is updated to use a boolean, rather that the `:` token (or a synthetic token), to determine whether initializers should be processed as part of building the outline. The parsing of initializers is done using the `:` token, if any, now pass both from regular constructors and primary constructor body declarations.
Closes#63468
Change-Id: Ia266dea7946b30b02d56f9f3cedf40848c4ee440
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/507401
Reviewed-by: Konstantin Shcheglov <scheglov@google.com>
Danny Tuppeny