Some tests in pkg/frontend/test are actually test suites written using the
pkg/testing framework.
This change runs these suites using a custom test runner instead of simply
running the test file, which makes the individual test results, timings and
logs available to the test system.
Change-Id: Iae5f37d9bd1b3e9b26c66443fe10e5cca6a6b6bf
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/118540
Commit-Queue: Karl Klose <karlklose@google.com>
Reviewed-by: Jonas Termansen <sortie@google.com>
Reviewed-by: Johnni Winther <johnniwinther@google.com>
Currently Dart SDK has 2 platform dill files:
* vm_platform_strong.dill is used when compiling Dart sources (in kernel
service and various kernel compilers).
* vm_platform_strong_stripped.dill is used to build core snapshot,
so its contents is used for execution in the VM.
Before this change, if Dart SDK is built with bytecode, then both
vm_platform_strong.dill and vm_platform_strong_stripped.dill contain
bytecode and AST.
This change removes bytecode from vm_platform_strong.dill, and removes
AST from vm_platform_strong_stripped.dill.
Sizes:
Dart SDK is built without bytecode:
5861400 out/ReleaseX64/vm_platform_strong.dill
2819336 out/ReleaseX64/vm_platform_strong_stripped.dill
32105720 out/ReleaseX64/dart-sdk/bin/dart
Dart SDK is built with bytecode, before this change:
11146480 out/ReleaseX64/vm_platform_strong.dill
4846488 out/ReleaseX64/vm_platform_strong_stripped.dill
34219256 out/ReleaseX64/dart-sdk/bin/dart
Dart SDK is built with bytecode, after this change:
5861400 out/ReleaseX64/vm_platform_strong.dill
2027224 out/ReleaseX64/vm_platform_strong_stripped.dill
28382456 out/ReleaseX64/dart-sdk/bin/dart
Change-Id: Ia7c61a9bce1c95edfd3b2810a67c0964fb37377f
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/118371
Reviewed-by: Régis Crelier <regis@google.com>
Commit-Queue: Alexander Markov <alexmarkov@google.com>
This CL gives the option to load classes lazily (on by default),
meaning that fields, procedures, constructors and redirecting
factory constructors of classes are not initially read,
but instead read when (or if) needed.
The idea being that many classes aren’t needed, and that
spending time on reading them is thus just a waste of time.
This is especially true in modular compilation where a lot
of modules needs to be given as input as the compilation at
hand might technically rely on it, but in practice only very
few of the classes are actually needed (or very few classes
actually need their members).
The below numbers are comparing a VM without this change and without
the lazy class hierarchy member lists change with
b31566b297 (the base change for those
two changes).
For running a simple hello-world script, on my machine 500 times
and doing statistics on it I can measure the following differences:
With verbose, from kernel_service startup to compile complete:
Difference at 95.0% confidence
-17.8 +/- 0.555817
-24.3455% +/- 0.760206%
(Student's t, pooled s = 4.48379)
With verbose, total runtime measure by `time`
Difference at 95.0% confidence
-19.346 +/- 0.905217
-12.0333% +/- 0.563051%
(Student's t, pooled s = 7.30242)
Without verbose, total runtime measure by `time`
Difference at 95.0% confidence
-17.862 +/- 0.842905
-12.0635% +/- 0.569276%
(Student's t, pooled s = 6.79974)
For running a number of modular compilations I’ve recorded an
improvement in actual runtime in the order of 20-37% depending
on the circumstance. In a real build system, though, the
difference seems to be more in the order of around 5%.
Change-Id: Id329bcf2b01d12c12d7a49f2b8abacd9c2447f05
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/115703
Commit-Queue: Jens Johansen <jensj@google.com>
Reviewed-by: Johnni Winther <johnniwinther@google.com>
That should allow us to "implement"
6f85cb83cf again in two steps:
* first we remove the dart:profiler as a required library, but retain it in the platform
* (we wait for stuff to roll, both in flutter and in google3)
* we remove it completely (i.e. not retaining it in the platform, removing the source etc)
Change-Id: Ieb77c2ba909300c05f3923c2edba7cab7d34fae6
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/115244
Commit-Queue: Jens Johansen <jensj@google.com>
Reviewed-by: Johnni Winther <johnniwinther@google.com>
Output from a comparison could be something like
Integers differ: 2 vs 3
Map with key 'nullability' has different values (tag 95, likely 'TypeParameterType')
Lists have different values at index 0
Map with key 'items' has different values
Map with key 'typeArguments' has different values (tag 93, likely 'InterfaceType')
Map with key 'returnType' has different values (tag 3, likely 'FunctionNode')
Map with key 'value' has different values (tag 1, likely 'Something<1>')
Map with key 'function' has different values (tag 6, likely 'Procedure')
Lists have different values at index 0
Map with key 'items' has different values
Map with key 'procedures' has different values (tag 2, likely 'Class')
Lists have different values at index 6
Map with key 'items' has different values
Map with key 'classes' has different values
Lists have different values at index 458
Map with key 'libraries' has different values
Which should be read bottom up as something like
Component -> the 458th library -> the 6th class ->
the 0th procedure -> function -> returnType (a InterfaceType) ->
0th type argument (a TypeParameterType) ->
nullability has value 2 in one input and 3 in the other.
Change-Id: I49007a80ddb53a0ba1fae04b22d6efeae3687724
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/114602
Commit-Queue: Jens Johansen <jensj@google.com>
Reviewed-by: Johnni Winther <johnniwinther@google.com>
* Include source positions, source text (if provided by front-end, which is
controlled by a separate option) and assertions into platform with bytecode.
Also use environment defines if provided.
* Omit bytecode for synthetic static fields '_redirecting#' injected
by front-end, as they contain invalid AST nodes (taking tear-off of
a constructor). This is needed to avoid emitting bytecode which cannot be
loaded.
* Correctly merge source information such as source text and line numbers
in bytecode generator if the script was originally created for an
anonymous mixin application without such information.
* Always include import URIs into source information, even if other information is
excluded (for members in anonymous mixin applications / noSuchMethod forwarders /
forwarding stubs).
* Set native resolvers for vm/cc/PrintJSON test, as printing all objects to JSON
involves collecting token positions, which may read bytecode from platform and
need native resolvers to process native method entries.
Change-Id: I971e880c25439c0ec69db3a231a0085ccd5d15ad
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/113780
Commit-Queue: Alexander Markov <alexmarkov@google.com>
Reviewed-by: Régis Crelier <regis@google.com>
Runs a number of tests that are quick, currently:
* Messages test in a new fast mode which skips compilation
* The spelling tests
* The custom CFE lint test
Change-Id: I7b444e5be1cf2397d6d776aac96dd7c2ad2b302d
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/113320
Reviewed-by: Johnni Winther <johnniwinther@google.com>
Commit-Queue: Jens Johansen <jensj@google.com>
Users of the CFE often need the ClassHierarchy and may CoreTypes after
invoking the CFE. This CL extends the result of `kernelForProgram` and
`kernelForComponent` to a `CompilerResult` that includes these objects.
Change-Id: I08491198fe876b6514e49993ad794388d6662c9c
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/112259
Commit-Queue: Johnni Winther <johnniwinther@google.com>
Reviewed-by: Alexander Markov <alexmarkov@google.com>
This CL adds a simple spell-checker, adds a spell-checking-phase to
messages_test to spell-check our messages, adds a new test that
spell-check our dart-code and fixes a lot of spelling mistakes / typos.
Change-Id: I8943a5bd67970e9a8a775251ae0aa97799eab097
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/112346
Commit-Queue: Jens Johansen <jensj@google.com>
Reviewed-by: Johnni Winther <johnniwinther@google.com>
The many different entrypoints to the CFE have repeatedly shown problems
with targets that do not support unevaluated constants being used
without providing an environment.
Making the targets opt-in to supporting unevaluated constants will
give the users of CFE early notice that CFE is used inconsistently
(as opposed to the backends failing when they happen to see an
UnevaluatedConstant node that they don't support).
Change-Id: I98c80df3a551823598e413e2895f5649f22f7c22
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/110561
Reviewed-by: Vijay Menon <vsm@google.com>
Commit-Queue: Johnni Winther <johnniwinther@google.com>
During the outline phase, after top-level type inference, compile
annotations for libraries, classes, fields, procedures, and
constructors.
Change-Id: I95ca65fd58ad88d9452a28d5a0652bee44aeda3a
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/103806
Commit-Queue: Kevin Millikin <kmillikin@google.com>
Reviewed-by: Dmitry Stefantsov <dmitryas@google.com>
The only remaining code in perf.dart that was using the legacy
analyzer parser was in parseDirectives, which is part of setup, and
not part of the code being performance tested, so there should be no
effect on the benchmark.
Change-Id: I4676b10c367d477499c96f5220e622bb390752a4
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/100680
Reviewed-by: Samuel Rawlins <srawlins@google.com>
Commit-Queue: Paul Berry <paulberry@google.com>
Since Fasta is only concerned with compile-time behavior, the option
only affects asserts in initializers in const constructors when
evaluated during constant evaluation.
Change-Id: I534cc7cc1f33f9b34d15550d38ea930ed7ac6709
Reviewed-on: https://dart-review.googlesource.com/c/94740
Commit-Queue: Aske Simon Christensen <askesc@google.com>
Commit-Queue: Peter von der Ahé <ahe@google.com>
Reviewed-by: Peter von der Ahé <ahe@google.com>
Auto-Submit: Aske Simon Christensen <askesc@google.com>
Jens Johansen