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.
Quick quiz:
Q: In a CFG, what does `a -> b -> c -> d` mean?
A: `a`, then `b`, then `c`, then `d`.
Q: In a CFG, what does `a -> b -> d; a -> c -> d` mean?
A: `a`, then `b` or `c`, then `d`.
Q: So how do you encode "a, then (b, then c) or (c, then b), then d`?
A: You can't.
Problem is, you need to, because that's what `a; run2({ b }, { c }); d`
does when `run2` has a contract that it calls both its lambda arguments
in-place: `shuffle(listOf(block1, block2)).forEach { it() }` is a
perfectly valid implementation for it, as little sense as that makes.
So that's what union nodes solve. When a node implements
`UnionNodeMarker`, its inputs are interpreted as "all visited in some
order" instead of the normal "one of the inputs is visited".
Currently this is used for data flow. It *should* also be used for
control flow, but it isn't. But it should be. But that's not so easy.
BTW, `try` exit is NOT a union node; although lambdas in one branch can
be completed according to types' of lambdas in another, data does not
flow between the branches anyway (since we don't know how much of the
`try` executed before jumping into `catch`, and `catch`es are mutually
exclusive) so a `try` expression is more like `when` than a function
call with called-in-place-exactly-once arguments. The fact that
`exitTryExpression` used `processUnionOfArguments` in a weird way
should've hinted at that, but now we know for certain.
It's also not a backwards jump in do-while, unless it's in the loop's
condition, which is a stupid "feature" IMO. As you can probably tell
from the comments added in this commit.
Make smart-casts non-transparent expression without delegation
to underlying FirQualifiedAccessExpression, as children delegation in
fir tree has unclear semantics
Remove two different kinds of tree nodes for smart-casts
* Change 1.6 to 1.7 constants
* Fix SAFE_CALL_WILL_CHANGE_NULLABILITY for testData
* Change EXPOSED_PROPERTY_TYPE_IN_CONSTRUCTOR_WARNING to EXPOSED_PROPERTY_TYPE_IN_CONSTRUCTOR_ERROR
* Change NON_EXHAUSTIVE_WHEN_STATEMENT to NO_ELSE_IN_WHEN
* Fix testData for SafeCallsAreAlwaysNullable
* Change T -> T & Any in test dumps
* Change INVALID_CHARACTERS_NATIVE_WARNING -> INVALID_CHARACTERS_NATIVE_ERROR
* TYPECHECKER_HAS_RUN_INTO_RECURSIVE_PROBLEM_WARNING -> TYPECHECKER_HAS_RUN_INTO_RECURSIVE_PROBLEM_ERROR
Consider the following code:
```
fun test(a: List<String>) {
a.first()
}
```
The dispatch receiver type of `first` in this case is `List<T>` before
this change. After this change, it's `List<String>`.
In addition, this change also replace the dispatch receiver type with
the more specific type if available. For example, consider the following
```
class MyList: ArrayList<String>()
fun test(a: MyList) {
a.get(0)
}
```
The dispatch receiver type of `get` is `MyList`, instead of
`ArrayList<String>`. That is, a fake override is created in this case.
In order to make resolution still work for members not available from
`Nothing`, we track the type without `Nothing?` and use that for
resolution instead.
Previously unsafe call is reported as part of InapplicableWrongReceiver.
This makes it difficult for the downstream checkers to report different
diagnostics.
To do so, inside the root cause of inapplicable candidate errors,
we will record expected/actual type of receiver, if any.
That will help identifying inapplicable calls on nullable receiver.
Update includes:
- Changing syntax of `OI/`NI` tags from `<!NI;TAG!>` to `<!TAG{NI}!>`
- Fix some incorrect directives
- Change order of diagnostics in some places
- Remove ignored diagnostics from FIR test data (previously `DIAGNOSTICS` didn't work)
- Update FIR dumps in some places and add `FIR_IDENTICAL` if needed
- Replace all JAVAC_SKIP with SKIP_JAVAC directive
Before that commit we desugared `a ?: b` as
when (val elvis = a) {
null -> b
else -> elvis
}
It was incorrect, because `a` should be resolved in dependent mode,
but when it was `elvis` initializer it was resolved in independent
mode, so we can't infer type for `a` in some complex cases
Denis.Zharkov