Previously the graph was constructed in such a way that
:function_type_arguments_var was only initialized when we call the
function with yield points for the first time. Which meant that on
resumption we would the type arguments.
Alternative to this fix would be to capture :function_type_arguments_var
but that does not make sense: yield points only occur inside
non-generic functions at the moment, which means that :function_type_arguments_var
is cheaper then capturing it as it is just a load of a field
from the closure object.
Fixes https://github.com/dart-lang/sdk/issues/33025
Bug: 33025
Change-Id: I0543b64202511a6c48744d462e384cf8a79e903e
Reviewed-on: https://dart-review.googlesource.com/53664
Commit-Queue: Vyacheslav Egorov <vegorov@google.com>
Reviewed-by: Martin Kustermann <kustermann@google.com>
Also fix the type inference algorithm so that when it explores
multiple alternatives (as a result of the use of FutureOr<>), if one
alternative generated some constraints but failed to produce a match,
it rewinds those constraints and tries the other alternative.
(Previously, it erroneously assumed that if any constraints were
generated, the alternative matched successfully).
Also add unit tests to verify the proper operation of the subtype
match algorithm.
Fixes#32305.
Change-Id: I060b5d6d5247a68d2b27bba78819bae172e43d97
Reviewed-on: https://dart-review.googlesource.com/53685
Commit-Queue: Paul Berry <paulberry@google.com>
Reviewed-by: Jenny Messerly <jmesserly@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>
Since there were already non-negative tests with the same numbers, I
gave them meaningful names.
Merged a couple of tests into the same file when it made sense since
multitests let us do that.
Deleted prefix1_negative_test, because it essentially tested only that
a random identifier does not resolve, which is not particularly related
to prefixes.
Change-Id: I91718f6df5126f123d9a8ea3abbc6898797db2db
Reviewed-on: https://dart-review.googlesource.com/52985
Reviewed-by: Sigmund Cherem <sigmund@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>
Found two mistakes in the tests:
- label2_negative_test had no labels whatsoever.
- label8_negative_test called an undefined function, doAgain().
Change-Id: I4d17c969a2b1422ff96cf68996225ecaaf5fe829
Reviewed-on: https://dart-review.googlesource.com/52868
Commit-Queue: Bob Nystrom <rnystrom@google.com>
Reviewed-by: Sigmund Cherem <sigmund@google.com>
In the process, I discovered multitests can't handle imports to
non-existent libraries. So I fixed that and cleaned up the code.
Change-Id: I9a8557e84f91ba7858bdf98f8732cd0ded55aa1a
Reviewed-on: https://dart-review.googlesource.com/52869
Commit-Queue: Bob Nystrom <rnystrom@google.com>
Reviewed-by: Leaf Petersen <leafp@google.com>
Most notably this fixes mixins, but a few others issues were discovered
as well. Fixes#32428.
This gets dartdevk closer to producing a valid Dart SDK file
Change-Id: I2973baef279d6b71ed29b97bec758b2d3209c275
Reviewed-on: https://dart-review.googlesource.com/51760
Commit-Queue: Jenny Messerly <jmesserly@google.com>
Reviewed-by: Vijay Menon <vsm@google.com>
The values returned from calling sync*, async and async* are now parameterised.
Simple cases where the type has no type parameters are handled.
The next step is to introduce a signature-like j-model entity to compute the type in more complex cases.
Change-Id: I2e43088cc6383a70e3a3d3067ec7542694ee2d0a
Reviewed-on: https://dart-review.googlesource.com/51421
Commit-Queue: Stephen Adams <sra@google.com>
Reviewed-by: Sigmund Cherem <sigmund@google.com>
One thing to be careful of here is that function type bounds may be
left unbounded. However, their parameters and return type cannot refer
to a type parameter for resolve-to-bounds to work.
There are also a large number of ways that this can loop infinitely, and
a number of new guards had to be added (as well as some test cases) to
catch them all.
Change-Id: I14322f5f71d1f7b73b27a870553e5c588b2c7e2e
Reviewed-on: https://dart-review.googlesource.com/52062
Commit-Queue: Mike Fairhurst <mfairhurst@google.com>
Reviewed-by: Konstantin Shcheglov <scheglov@google.com>
All fixes are related to the requirement that a concrete class must
implement its entire interface or have a non-trivial noSuchMethod
implementation.
Change-Id: I015e0565e5dd9cfa324c38bb43696e365c58329a
Reviewed-on: https://dart-review.googlesource.com/51880
Commit-Queue: Aske Simon Christensen <askesc@google.com>
Reviewed-by: Dmitry Stefantsov <dmitryas@google.com>
Jenny Messerly