This is something we'll want to investigate. It looks like dartdevk's
performance using its own SDK is slightly worse, suggesting it's missing
some optimizations (e.g. `JS` types that are not recognized, extra null
checks, covariance check elimination).
Change-Id: Ieaedd6be3bfb70718f495fc86cb1adcf67aad3cc
Reviewed-on: https://dart-review.googlesource.com/59580
Reviewed-by: Jenny Messerly <jmesserly@google.com>
Currently we test dartdevk against SDK/packages that were compiled with
dartdevc (DDC+Analyzer). This CL migrates to having those files compiled
with DDC+Kernel. dartdevc testing is unchanged.
Most of the fixes are around things like special optimized annotations
used in our SDK code, that were not understood by DDK. Also some inline
JS type annoations were not correct.
Change-Id: Iccf4427e4b9beffd6d97a4be654253d91f6cb89e
Reviewed-on: https://dart-review.googlesource.com/57800
Reviewed-by: Vijay Menon <vsm@google.com>
Dynamic invocations of generic functions might not pass type arguments.
The type arguments need to be filled in with the default type arguments
calculated by the CFE.
Change-Id: I5865e79b2975af97d1ca1680bd88ba8137208bb8
Reviewed-on: https://dart-review.googlesource.com/48423
Commit-Queue: Samir Jindel <sjindel@google.com>
Reviewed-by: Régis Crelier <regis@google.com>
The behavioral difference is that named and optional arguments are filled in
with their default values in the `Invocation` object passed to `noSuchMethod`.
On the implementation side we make NSM forwarders concrete and fill in their
bodies in the CFE. The custom (and somewhat hacky) VM support is no longer
needed, and Dart2JS can benefit from this implementation as well.
According to discussion on #33031 we will be able to re-land this soon without
breaking Mockito.
Prior failures on precompiler bots are fixed in Patchset 2.
Change-Id: If1b7fe4cf6da5ef38f330e1ad226121bcfc958a1
Reviewed-on: https://dart-review.googlesource.com/54401
Commit-Queue: Samir Jindel <sjindel@google.com>
Reviewed-by: Dmitry Stefantsov <dmitryas@google.com>
sync*/async/async* generators have two parts: an entry, which does
checks and computes the element type of the result, and a body, which
is transformed after lowering to JavaScript.
The split ensures that the code for checks is done at function call
time, rather than e.g. in the moveNext method of a sync* iterator.
A following CL with optimize the split to generate one function when
there are no checks and the type argument is simple enough for textual
substitution.
Change-Id: I5414109ca851f9267871aa113a2e29b16236986d
Reviewed-on: https://dart-review.googlesource.com/54308
Commit-Queue: Stephen Adams <sra@google.com>
Reviewed-by: Sigmund Cherem <sigmund@google.com>
Reviewed-by: Johnni Winther <johnniwinther@google.com>
This reverts commit cf78da8a48.
Reason for revert: introduces significant performance regression (~30%) on analyzer benchmarks (warm-analysis) without clearly visible hot-spot.
Original change's description:
> [VM] Reduce Smi size to 32 bit on 64 bit platforms
>
> This reduces small tagged integers on 64 bit platforms from 63 bits to
> 31 bits plus one tag bit.
> This is a step on the way to compile-time-optional compressed pointers
> on 64 bit platforms. See more about this at go/dartvmlearnings
> This causes a slowdown for some uses of integers that don't fit in 31
> signed bits, but because both x64 and ARM64 have unboxed 64 bit
> integers now the performance hit should not be too bad.
>
> This is a reapplication of
> https://dart-review.googlesource.com/c/sdk/+/46244
> It was reverted due to a compilation error on 32 bit
> ARM with DBC.
>
> R=vegorov@google.com
>
> Change-Id: I943de1768519457f0e5a61ef0b4ef204b6a53281
> Reviewed-on: https://dart-review.googlesource.com/51321
> Reviewed-by: Vyacheslav Egorov <vegorov@google.com>
TBR=vegorov@google.com,erikcorry@google.com
# Not skipping CQ checks because original CL landed > 1 day ago.
Change-Id: I8c5b909ec38663b5f5b05f69ef488c97341f8f3d
Reviewed-on: https://dart-review.googlesource.com/54000
Reviewed-by: Vyacheslav Egorov <vegorov@google.com>
Commit-Queue: Vyacheslav Egorov <vegorov@google.com>
This reverts commit 7d5025e814.
Reason for revert: <INSERT REASONING HERE>
Original change's description:
> Fix incorrect handling of NSM forwarders and pull all logic into CFE.
>
> The original implementation was designed around a shared misunderstanding
> of optional parameter handling in the spec. (which was also ambiguous about type parameters).
>
> The correct behavior for optional/type parameters is to fill them in with their default values/bounds.
> This behavior can be implemented without any backend support, as is done in this CL.
>
> Change-Id: Ib81f17ead2d2920e755703e244db5afc328d0315
> Reviewed-on: https://dart-review.googlesource.com/52802
> Commit-Queue: Samir Jindel <sjindel@google.com>
> Reviewed-by: Dmitry Stefantsov <dmitryas@google.com>
TBR=dmitryas@google.com,sjindel@google.com
Change-Id: Ice6cd36cc62772e013bded83e0f589fe4b5e9d53
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://dart-review.googlesource.com/53400
Reviewed-by: Samir Jindel <sjindel@google.com>
Commit-Queue: Samir Jindel <sjindel@google.com>
The original implementation was designed around a shared misunderstanding
of optional parameter handling in the spec. (which was also ambiguous about type parameters).
The correct behavior for optional/type parameters is to fill them in with their default values/bounds.
This behavior can be implemented without any backend support, as is done in this CL.
Change-Id: Ib81f17ead2d2920e755703e244db5afc328d0315
Reviewed-on: https://dart-review.googlesource.com/52802
Commit-Queue: Samir Jindel <sjindel@google.com>
Reviewed-by: Dmitry Stefantsov <dmitryas@google.com>
This reduces small tagged integers on 64 bit platforms from 63 bits to
31 bits plus one tag bit.
This is a step on the way to compile-time-optional compressed pointers
on 64 bit platforms. See more about this at go/dartvmlearnings
This causes a slowdown for some uses of integers that don't fit in 31
signed bits, but because both x64 and ARM64 have unboxed 64 bit
integers now the performance hit should not be too bad.
This is a reapplication of
https://dart-review.googlesource.com/c/sdk/+/46244
It was reverted due to a compilation error on 32 bit
ARM with DBC.
R=vegorov@google.com
Change-Id: I943de1768519457f0e5a61ef0b4ef204b6a53281
Reviewed-on: https://dart-review.googlesource.com/51321
Reviewed-by: Vyacheslav Egorov <vegorov@google.com>
Type arguments that require a subsitution need the interceptor to
access the substitution function.
- Add an interceptor input to HTypeInfoReadVariable
- Add an entry point getRuntimeTypeArgumentIntercepted that takes the interceptor.
- Optimize uses that do not need the interceptor.
The interceptor is cheap to get from the context in the common case
that the type variable access is directly in an instance method.
A typical use (broken)
var t1 = H.getRuntimeTypeArgument(receiver, "ListMixin", 0);
becomes
var t1 = H.getRuntimeTypeArgumentIntercepted(this, receiver, "ListMixin", 0);
Change-Id: I2abe6a2b6e8d2cdb0ede1b7248dd8337830f8a0c
Reviewed-on: https://dart-review.googlesource.com/50640
Commit-Queue: Stephen Adams <sra@google.com>
Reviewed-by: Johnni Winther <johnniwinther@google.com>
Reviewed-by: Sigmund Cherem <sigmund@google.com>
* Preserve empty lines in the file header.
* Add empty entry to the tool's newly created sections.
* Remove extra empty line at the end of the file.
Change-Id: I271583774d5f5497025a9d85bcadf8b0b9e39e81
Reviewed-on: https://dart-review.googlesource.com/37600
Commit-Queue: Alexander Thomas <athom@google.com>
Reviewed-by: William Hesse <whesse@google.com>
Régis Crelier