Previously, a function reference that used generic parameters from its
outer scope was lowered into a top-level non-generic subclass of
`FunctionN`, with `FunctionN` type arguments referencing type parameters
not present in the scope anymore. This sometimes resulted in generating
malformed mangled names.
From now on the generated subclass of `FunctionN` is generic. The needed
type arguments are passed upon instantiation, when the relevant generic
parameters are present in the scope.
function enter -> default 1 -> default 2 -> rest of function
\----------^ \----------^
This probably has no effect (in non-stupid code, at least), but it makes
graph construction more architecturally correct (now value parameters'
subgraphs get attached to a node).
Interpretation: a graph A is a subgraph of B if information available at
nodes of A depends on the paths taken in B. For example, local classes
are subgraphs of a graph in which they are declared, and members of
those classes are subgraphs of the local class itself - because these
members can reference captured values.
Consequences:
* if graph G is a subgraph of node N, then G is a subgraph of N's
owner;
* `ControlFlowAnalysisDiagnosticComponent` will only visit root graphs;
* `graph.traverse` will ignore subgraph boundaries, as if all subgraphs
are inlined into one huge root graph;
* if a control flow checker needs information from a declaration to
which a graph is attached, it must look at subgraphs explicitly.
For example, consider the `callsInPlace` checker. When a function
has a `callsInPlace` contract and a local declaration, the checker must
visit that local declaration to ensure it does not capture the allegedly
called-in-place argument - hence `graph.traverse` will look at the
nodes. However, the local declaration can also be a function with its
own `callsInPlace` contracts, so the checker should also run for it in
isolation. If that sounds quadratic, that's because unfortunately it is.
- If too few or too many type arguments were provided, they were all
thrown away in `TypeArgumentMapping`,
`FirCallCompletionResultsWriterTransformer`, and `KtFirCallResolver`.
The fix handles type arguments of the wrong arity more gracefully.
- Note for `TypeArgumentMapping`: Excess type arguments are not needed
for candidate resolution. Excess type arguments are still resolved
due to the handling in `FirCallCompletionResultsWriterTransformer`.
- Post-processing in `AllCandidatesResolver`: When all candidates are
resolved in `AllCandidatesResolver.getAllCandidates`, the function
builds a FIR file. During that resolution, the
`generic<String, String>` call (in example
`functionCallWithTooFewTypeArguments.kt`) is correctly marked as
inapplicable, but the missing type argument is inferred as an error
type. `firFile` then contains a function call
`generic<String, String, ERROR>` instead of `generic<String, String>`.
This call is still marked as inapplicable. Despite that, the
*subsequent* resolution by
`bodyResolveComponents.callResolve.collectAllCandidates` disregards
the call's inapplicability and resolves successfully into an
applicable candidate. This is because `CandidateFactory` doesn't make
any guarantees for already inapplicable calls. The fix adds
post-processing to `AllCandidatesResolver` to preserve candidate
inapplicability.
- Most tests that this commit changes had slightly different results due
to type arguments becoming resolvable.
- `wrongNumberOfTypeArguments.kt` and
`wrongNumberOfArgumentsInTypeAliasConstructor.kt`:
`ConeDiagnostic.toFirDiagnostics` prefers specific errors. Because
`ARGUMENT_TYPE_MISMATCH` is specific and `INAPPLICABLE_CANDIDATE` is
not, only the former is reported. I see no reason to pass an illegally
typed argument in either test, so the change reduces the errors to
`INAPPLICABLE_CANDIDATE`.
- `typeAliasSamAdapterConstructors2.fir.kt`: See KT-55007.
- Disable `mismatchTypeParameters` JS backend test due to its handling
of excess type arguments. See KT-55250.
^KT-54980 fixed
Newly added tests are basically copies of the existing tests on `until`.
Note that this operator is optimized for all backends, but the fact that
it's optimized is only checked for the JVM backend in bytecode text
tests.
#KT-53330 Fixed
Vladimir Sukharev