We introduce a class KernelSyntheticExpression which can be used for
synthetic expressions introduced by the front end. It is used for
expressions that represent errors detected by the front end, and it is
also the base class for KernelComplexAssignment. By default, type
inference skips these synthetic expressions, inferring a type of
`dynamic`. (Note, however, that classes dervied from
KernelComplexAssignment override this behavior).
We also introduce a class KernelSyntheticStatement which behaves
similarly, but is used for statements rather than expressions,
e.g. the synthetic assignment introduced at the top of a "for" loop
whose loop variable is defined elsewhere.
Finally, we modify KernelTypeInferrer.inferFieldTopLevel() so that it
deliberatly infers `dynamic` for fields that lack an initializer.
Previously this behavior happened as a result of the hack in
KernelTypeInferrer.inferExpression which caused type inference to
infer `dynamic` for anything that didn't implement KernelExpression.
All of these changes should help pave the way for removing the
aforementioned hack.
R=scheglov@google.com
Review-Url: https://codereview.chromium.org/2968563002 .
This involves a few pieces:
- Add support to DDC for specifying the module name associated with a
given summary. This lets test.dart invoke DDC using summaries in the
build directory outside of the directory containing the test itself.
- Add support to the build scripts for building the packages. This adds
a new GN target that builds everything needed to run test.dart with
dartdevc. In particular, it invokes build_pkgs.dart to compile the
relevant packages to JS+summary so that the tests can use them.
This requires some changes to build_pkgs.dart so it can output to a
given directory.
- In test.dart, when tests are compiled with dartdevc, pass in the
summaries for the packages so they don't get compiled in. Then, when
the test is run, configure require.js with the right paths to their
JS files so they can be loaded.
I also removed a bunch of unneeded buildDir parameters being passed
around the various CompilerConfiguration class methods now that they
have direct access to the configuration.
Fix#29923.
R=vsm@google.com, whesse@google.com, zra@google.com
Review-Url: https://codereview.chromium.org/2955513002 .
This makes 'implemented' notification 10x faster for files with
many types and implemented methods, like visitor.dart - from 400 ms
down to 36 ms. Still not as fast as I would like, but much better.
R=brianwilkerson@google.com
BUG=
Review-Url: https://codereview.chromium.org/2955313003 .
In general case we cannot use getResult() in this notification.
We generate these notifications after analysis is complete in scheduler.
But if there is no cached result, getResult() will schedule analysis.
So, we might get into a loop.
In practice this worked, because IDEA always makes files priority,
so their results cached, and then subscribes for notifications.
But in general case we need this protection.
This also means that the 'implemented' notification works (and will
work) only for priority files.
R=brianwilkerson@google.com
BUG=
Review-Url: https://codereview.chromium.org/2965533002 .
This prepares for using LocalFunctionElement.callMethod as the key for local
functions instead of the local functions themselves - corresponding to the world
after closure conversion.
Main changes in
elements/elements.dart
inferrer/inferrer_engine.dart
inferrer/type_graph_inferrer.dart
inferrer/type_system.dart
R=sigmund@google.com
Review-Url: https://codereview.chromium.org/2955353002 .
This is intended for profiling, to ensure that type inference and type
promotion do not slow down compilation too much.
Currently the option is just plumbed through to the kernel Target
class since I'm not exactly sure how we'll want to enable/disable this
option; we can plumb it the rest of the way in a future CL.
R=ahe@google.com
Review-Url: https://codereview.chromium.org/2962913002 .
Currently some tests pass in checked mode but fails in non-checked mode
because we don't check the result of a constant-evaluation that should
be bool in non-checked mode.
This CL changes it so we always check the bool result in the constant-evaluator.
BUG=
R=kustermann@google.com
Review-Url: https://codereview.chromium.org/2956493003 .
_ChildNodeListLazy.removeRange used to throw UnsupportedError because ListMixin.removeRange calls setRange.
Make throwing UnsupportedError explicit to avoid relying on the implementation of ListMixin.removeRange
This will unblock the '_closeGap' change.
BUG=
R=lrn@google.com
Review-Url: https://codereview.chromium.org/2955103003 .
