| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The pattern matcher needs to substitute references to bound
variables with references to either a) synthetic temporary vals,
or to b) selections. The latter occurs under -optimize to avoid
to be frugal with local variable slots.
For instance:
```
def test(s: Some[String]) = s match {
case Some(elem) => elem.length
}
```
Is translated to:
```
def test(s: Some[String]): Int = {
case <synthetic> val x1: Some[String] = s;
case4(){
if (x1.ne(null))
matchEnd3(x1.x.length())
else
case5()
};
case5(){
matchEnd3(throw new MatchError(x1))
};
matchEnd3(x: Int){
x
}
}
```
However, for a long time this translation failed to consider
references to the binder in types. #4122 tried to address this
by either using standard substitution facilities where available
(references to temp vals), and by expanding the patmat's
home grown substitution to handle the more complex case of
referencing a selection.
However, this left the tree in an incoherent state; while it
patched up the `.tpe` field of `Tree`s, it failed to modify the
info of `Symbol`-s.
This led to a crash in the later uncurry phase under
`-Ydelambdafy:method`.
This commit modifies the info of such symbols to get rid of stray
refeferences to the pattern binder symbols.
|
| |\
| |
| | |
SI-5154 Parse leading literal brace in XML pattern
|
| | |
| |
| |
| |
| |
| |
| | |
Don't consume literal brace as Scala pattern.
Previously, leading space would let the text parser `xText`
handle it correctly instead.
|
| |\ \
| | |
| | | |
SI-9087 Fix min/max of reversed Double/Float orderings
|
| | | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
As diagnosed by the reporter, we needed additional overrides
due to the way these orderings are implemented.
I've added tests to show other methods and other orderings
are working correctly.
After writing that, I found a scalacheck test related to
NaN handling that also covers `Ordering`. I had to correct
the assertion in the tests of `reverse.{min,max}`.
|
| | | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This infrastructure is required for the inliner: when inlining code
from a classfile, the corresponding ClassBType is needed for various
things (eg access checks, InnerClass attribute).
The test creates two ClassBTypes for the same class: once using the
(unpickled) Symbol, once using the parsed ASM ClassNode, and verifies
that the two are the same.
There's a cleanup to the InnerClass attribute:
object T { class Member; def foo = { class Local } }
class T
For Java compatibility the InnerClass entry for Member says the class
is nested in T (not in the module class T$). We now make sure to add
that entry only to T, not to T$ (unless Member is actually referenced
in the classfile T$, in that case it will be added, as required).
|
| | | |
| | |
| | |
| | |
| | | |
Introduces methods for textifying classes, methods, InsnLists and
individual AbstractInsnNodes.
|
| |\ \ \
| |/ /
|/| | |
SI-9089 Another REPL/FSC + specialization bug fix
|
| | | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
The enclosed test case stopped working in 2.11.5 on the back of
https://github.com/scala/scala/pull/4040.
The key change was that we ran all post-typer info transformers
on each run of the compiler, rather than trying to reuse the results
of the previous run.
In that patch, I noticed one place [1] in specialization that
aggressively entered specialized members into the owning scope,
rather than relying on `transformInfo` to place the new members
in the scope of the newly created element of the info history.
I made that change after noticing that this code could actually
mutated scopes of specializaed types at the parser phase, which
led to fairly obscure failures.
This bug is another one of these obscure failures, and has the
same root cause. We effectively "double specialiaze" Function0,
which trips an assertion when `method apply$mcI$sp` is found
twice in a scope.
I have found another spot that was directly manipulating the scope,
and removed the offending code.
[1] https://github.com/scala/scala/pull/4040#commitcomment-8531516
|
| |\ \ \
| |/ /
|/| | |
Fix many typos in docs and comments
|
| | |/
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This commit corrects many typos found in scaladocs, comments and
documentation. It should reduce a bit number of PRs which fix one
typo.
There are no changes in the 'real' code except one corrected name of
a JUnit test method and some error messages in exceptions. In the case
of typos in other method or field names etc., I just skipped them.
Obviously this commit doesn't fix all existing typos. I just generated
in IntelliJ the list of potential typos and looked through it quickly.
|
| |\ \
| | |
| | | |
SI-7965 Support calls to MethodHandle.{invoke,invokeExact}
|
| | | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
These methods are "signature polymorphic", which means that compiler
should not:
1. adapt the arguments to `Object`
2. wrap the repeated parameters in an array
3. adapt the result type to `Object`, but instead treat it as it
it already conforms to the expected type.
Dispiritingly, my initial attempt to implement this touched the type
checker, uncurry, erasure, and the backend.
However, I realized we could centralize handling of this in the typer
if at each application we substituted the signature polymorphic
symbol with a clone that carried its implied signature, which is
derived from the types of the arguments (typechecked without an
expected type) and position within and enclosing cast or block.
The test case requires Java 7+ to compile so is currently embedded
in a conditionally compiled block of code in a run test.
We ought to create a partest category for modern JVMs so we can
write such tests in a more natural style.
Here's how this looks in bytecode. Note the `bipush` / `istore`
before/after the invocation of `invokeExact`, and the descriptor
`(LO$;I)I`.
```
% cat sandbox/poly-sig.scala && qscala Test && echo ':javap Test$#main' | qscala
import java.lang.invoke._
object O {
def bar(x: Int): Int = -x
}
object Test {
def main(args: Array[String]): Unit = {
def lookup(name: String, params: Array[Class[_]], ret: Class[_]) = {
val lookup = java.lang.invoke.MethodHandles.lookup
val mt = MethodType.methodType(ret, params)
lookup.findVirtual(O.getClass, name, mt)
}
def lookupBar = lookup("bar", Array(classOf[Int]), classOf[Int])
val barResult: Int = lookupBar.invokeExact(O, 42)
()
}
}
scala> :javap Test$#main
public void main(java.lang.String[]);
descriptor: ([Ljava/lang/String;)V
flags: ACC_PUBLIC
Code:
stack=3, locals=3, args_size=2
0: aload_0
1: invokespecial #18 // Method lookupBar$1:()Ljava/lang/invoke/MethodHandle;
4: getstatic #23 // Field O$.MODULE$:LO$;
7: bipush 42
9: invokevirtual #29 // Method java/lang/invoke/MethodHandle.invokeExact:(LO$;I)I
12: istore_2
13: return
LocalVariableTable:
Start Length Slot Name Signature
0 14 0 this LTest$;
0 14 1 args [Ljava/lang/String;
13 0 2 barResult I
LineNumberTable:
line 16: 0
}
```
I've run this test across our active JVMs:
```
% for v in 1.6 1.7 1.8; do java_use $v; pt --terse test/files/run/t7965.scala || break; done
java version "1.6.0_65"
Java(TM) SE Runtime Environment (build 1.6.0_65-b14-466.1-11M4716)
Java HotSpot(TM) 64-Bit Server VM (build 20.65-b04-466.1, mixed mode)
Selected 1 tests drawn from specified tests
.
1/1 passed (elapsed time: 00:00:02)
Test Run PASSED
java version "1.7.0_71"
Java(TM) SE Runtime Environment (build 1.7.0_71-b14)
Java HotSpot(TM) 64-Bit Server VM (build 24.71-b01, mixed mode)
Selected 1 tests drawn from specified tests
.
1/1 passed (elapsed time: 00:00:07)
Test Run PASSED
java version "1.8.0_25"
Java(TM) SE Runtime Environment (build 1.8.0_25-b17)
Java HotSpot(TM) 64-Bit Server VM (build 25.25-b02, mixed mode)
Selected 1 tests drawn from specified tests
.
1/1 passed (elapsed time: 00:00:05)
Test Run PASSED
```
|
| |\ \ \
| | | |
| | | | |
SI-9044 Fix order of interfaces in classfiles
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
It was reversed since ced3ca8ae1. The reason is that the backend used
`mixinClasses` to obtain the parents of a class, which returns them in
linearization order.
`mixinClasses` als returns all ancestors (not only direct parents),
which means more work for `minimizeInterfaces`. This was introduced
in cd62f52 for unclear reasons. So we switch back to using `parents`.
|
| |\ \ \ \
| | | | |
| | | | | |
SI-8999 Reduce memory usage in exhaustivity check
|
| | | |_|/
| |/| |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
The OOM could occur when all models are forcibly expanded in the DPLL solver.
The simplest solution would be to limit the number of returned models but then
we are back in non-determinism land (since the subset we get back depends on
which models were found first).
A better alternative is to create only the models that have corresponding
counter examples.
If this does not ring a bell, here's a longer explanation:
TThe models we get from the DPLL solver need to be mapped back to counter
examples. However there's no precalculated mapping model -> counter example.
Even worse, not every valid model corresponds to a valid counter example.
The reason is that restricting the valid models further would for example
require a quadratic number of additional clauses. So to keep the optimistic case
fast (i.e., all cases are covered in a pattern match), the infeasible counter
examples are filtered later.
The DPLL procedure keeps the literals that do not contribute to the solution
unassigned, e.g., for `(a \/ b)` only {a = true} or {b = true} is required and
the other variable can have any value. This function does a smart expansion of
the model and avoids models that have conflicting mappings.
For example for in case of the given set of symbols (taken from `t7020.scala`):
"V2=2#16"
"V2=6#19"
"V2=5#18"
"V2=4#17"
"V2=7#20"
One possibility would be to group the symbols by domain but
this would only work for equality tests and would not be compatible
with type tests.
Another observation leads to a much simpler algorithm:
Only one of these symbols can be set to true,
since `V2` can at most be equal to one of {2,6,5,4,7}.
TODO: How does this fare with mixed TypeConst/ValueConst assignments?
When the assignments are e.g., V2=Int | V2=2 | V2=4, only the assignments
to value constants are mutually exclusive.
|
| |\ \ \ \
| | | | |
| | | | | |
SI-7459 Handle pattern binders used as prefixes in TypeTrees.
|
| | | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
Match translation was incorrect for:
case t => new t.C
case D(t) => new d.C
We would end up with Types in TypeTrees referring to the wrong
symbols, e.g:
// t7459a.scala
((x0$1: this.LM) => {
case <synthetic> val x1: this.LM = x0$1;
case4(){
matchEnd3(new tttt.Node[Any]())
};
matchEnd3(x: Any){
x
}
Or:
// t7459b.scala
((x0$1: CC) => {
case <synthetic> val x1: CC = x0$1;
case4(){
if (x1.ne(null))
matchEnd3(new tttt.Node[Any]())
else
case5()
};
This commit:
- Changes `bindSubPats` to traverse types, as well as terms,
in search of references to bound symbols
- Changes `Substitution` to reuse `Tree#substituteSymbols` rather
than the home-brew substitution from `Tree`s to `Tree`s, if the
`to` trees are all `Ident`s
- extends `substIdentsForTrees` to substitute selections like
`x1.caseField` into types.
I had to dance around the awkward handling of "swatches" (exception
handlers that can be implemented with JVM native type switches) by
duplicating trees to avoid seeing the results of `substituteSymbols`
in `caseDefs` after we abandon that approach if we detect the
patterns are too complex late in the game.
I also had to add an escape hatch for the "type selection from
volatile type" check in the type checker. Without this, the
translation of `pos/t7459c.scala`:
case <synthetic> val x1: _$1 = (null: Test.Mirror[_]).universe;
case5(){
if (x1.isInstanceOf[Test.JavaUniverse])
{
<synthetic> val x2: _$1 with Test.JavaUniverse = (x1.asInstanceOf[_$1 with Test.JavaUniverse]: _$1 with Test.JavaUniverse);
matchEnd4({
val ju1: Test.JavaUniverse = x2;
val f: () => x2.Type = (() => (null: x2.TypeTag[Nothing]).tpe);
.. triggers that error at `x2.TypeTag`.
|
| |\ \ \ \ \
| | | | | |
| | | | | | |
Suppress match analysis under -Xno-patmat-analysis
|
| | | |/ / /
| |/| | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
NoSuppression doesn't suppress. FullSuppression does.
Now using the latter when running under `-Xno-patmat-analysis`.
I should really have tested. /me hides under a rock
|
| |\ \ \ \ \
| |/ / / /
|/| | | | |
[nomerge] SI-9030 don't call private BoxesRunTime.equalsNumChar
|
| | | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
When comparing a Number and a Character, the would emit a call to the
private method. For binary compatibility, this method remains private
in 2.11, so we just use equalsNumObject instead.
|
| |\ \ \ \ \
| |_|_|/ /
|/| | | | |
Avoid the `CNF budget exceeded` exception via smarter translation into CNF.
|
| | | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
The exhaustivity checks in the pattern matcher build a propositional
formula that must be converted into conjunctive normal form (CNF) in
order to be amenable to the following DPLL decision procedure.
However, the simple conversion into CNF via negation normal form and
Shannon expansion that was used has exponential worst-case complexity
and thus even simple problems could become untractable.
A better approach is to use a transformation into an _equisatisfiable_
CNF-formula (by generating auxiliary variables) that runs with linear
complexity. The commonly known Tseitin transformation uses bi-
implication. I have choosen for an enhancement: the Plaisted
transformation which uses implication only, thus the resulting CNF
formula has (on average) only half of the clauses of a Tseitin
transformation.
The Plaisted transformation uses the polarities of sub-expressions to
figure out which part of the bi-implication can be omitted. However,
if all sub-expressions have positive polarity (e.g., after
transformation into negation normal form) then the conversion is
rather simple and the pseudo-normalization via NNF increases chances
only one side of the bi-implication is needed.
I implemented only optimizations that had a substantial
effect on formula size:
- formula simplification, extraction of multi argument operands
- if a formula is already in CNF then the Tseitin/Plaisted
transformation is omitted
- Plaisted via NNF
- omitted: (sharing of sub-formulas is also not implemented)
- omitted: (clause subsumption)
|
| |\ \ \ \ \
| | | | | |
| | | | | | |
SI-9015 Reject 0x and minor parser cleanup
|
| | | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
Only print error results. Show deprecated forms.
Test for rejected literals and clean up parser
There was no negative test for what constitutes a legal literal.
The ultimate goal is for the test to report all errors in
one compilation.
This commit follows up the removal of "1." syntax to simplify
number parsing. It removes previous paulp code to contain the
erstwhile complexity.
Leading zero is not immediately put to the buffer. Instead,
the empty buffer is handled on evaluation. In particular, an
empty buffer due to `0x` is a syntax error.
The message for obsolete octal syntax is nuanced and deferred
until evaluation by the parser, which is slightly simpler to
reason about.
Improve comment on usage of base
The slice-and-dicey usage of base deserves a better
comment. The difference is that `intVal` sees an empty
char buffer for input `"0"`.
|
| |\ \ \ \ \ \
| |_|_|/ / /
|/| | | | | |
SI-9008 Fix regression with higher kinded existentials
|
| | |/ / / /
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
Allow a naked type constructor in an existential type if
we are directly within a type application.
Recently, 84d4671 changed nested context creation to avoid passing
down the `TypeConstructorAllowed`, which led to missing kind errors
in code like `type T[({type M = List})#M]`.
However, when typechecking `T forSome { quantifiers }`, we create
a nested context to represent the nested scope introduced for the
quantifiers. But we need to propagate the `TypeConstructorAllowed`
bit to the nested context to allow for higher kinded existentials.
The enclosed tests show:
- pos/t9008 well kinded application of an hk existential
- neg/t9008 hk existential forbidden outside of type application
- neg/t9008b kind error reported for hk existential
Regressed in 84d4671.
|
| |\ \ \ \ \
| | | | | |
| | | | | | |
The alternative, flat representation of classpath elements
|
| | | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
This commit integrates with the compiler the whole flat classpath
representation build next to the recursive one as an alternative.
From now flat classpath really works and can be turned on. There's
added flag -YclasspathImpl with two options: recursive (the default
one) and flat.
It was needed to make the dynamic dispatch to the particular
classpath representation according to the chosen type of a classpath
representation.
There's added PathResolverFactory which is used instead of a concrete
implementation of a path resolver. It turned out that only a small
subset of path resolvers methods is used outside this class in Scala
sources. Therefore, PathResolverFactory returns an instance of a base
interface PathResolverResult providing only these used methods.
PathResolverFactory in combination with matches in some other places
ensures that in all places using classpath we create/get the proper
representation.
Also the classPath method in Global is modified to use the dynamic
dispatch. This is very important change as a return type changed to
the base ClassFileLookup providing subset of old ClassPath public
methods. It can be problematic if someone was using in his project
the explicit ClassPath type or public methods which are not provided
via ClassFileLookup. I tested flat classpath with sbt and Scala IDE
and there were no problems. Also was looking at sources of some other
projects like e.g. Scala plugin for IntelliJ and there shouldn't be
problems, I think, but it would be better to check these changes
using the community build.
Scalap's Main.scala is changed to be able to use both implementations
and also to use flags related to the classpath implementation.
The classpath invalidation is modified to work properly with the old
(recursive) classpath representation after changes made in a Global.
In the case of the attempt to use the invalidation for the flat cp it
just throws exception with a message that the flat one currently
doesn't support the invalidation. And also that's why the partest's
test for the invalidation has been changed to use (always) the old
implementation. There's added an adequate comment with TODO to this
file.
There's added partest test generating various dependencies
(directories, zips and jars with sources and class files) and testing
whether the compilation and further running an application works
correctly, when there are these various types of entries specified as
-classpath and -sourcepath. It should be a good approximation of real
use cases.
|
| | | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
The structure of scalap's Main has been refactored.
EmptyClasspath is deleted. It looks that it was unused since this
commit: https://github.com/scala/scala/commit/e594fe58ef8116a4bd2560ad0a856ad58ae9db33
Also classpath logging is changed and now uses asClassPathString
method. It was needed to modify one test because of that but it
won't depend on a particular representation.
There aren't changes in the way scalap works.
|
| |\ \ \ \ \ \
| | | | | | |
| | | | | | | |
SI-7683 Enable -Ystop-before:typer
|
| | | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
Allows a plugin to run before typer without incurring typechecking.
The test is that a plugin doesn't run when stopping after parser,
and that the truncated compilation run also succeeds, since
updating check files for the output of -Xshow-phases is tedious.
|
| |\ \ \ \ \ \ \
| | | | | | | |
| | | | | | | | |
SI-9027 Parser no longer consumes space after multi XML elements
|
| | |/ / / / / /
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
Once the parser starts looking for more <elements>, it should
still lookahead speculatively, leaving any remaining whitespace,
including newlines, after the last element.
|
| |\ \ \ \ \ \ \
| | | | | | | |
| | | | | | | | |
SI-9003 Eagerly capture more potentially mutable binders
|
| | | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | | |
This is a re-run of SI-5158 in two different contexts.
As reported, the result of `unapply[Seq]` under name based pattern
matching might not guarantee stability of results of calls to
`_1`, `apply(i)`, etc, so we call these eagerly, and call them
once only.
I also found a case in regular pattern matching that we hadn't
accounted for: extracting elements of sequences (either from
a case class or from an `unapplySeq`) may also be unstable.
This commit changes `ExtractorTreeMaker` to force storage
of such binders, even under `-optimize`. This parallels the change
to `ProductExtractorTreeMaker` in 8ebe8e3e8.
I have added a special case for traditional `unapply` methods
returning `Option`. This avoids a change for:
```
% cat test/files/run/t9003b.scala
object Single {
def unapply(a: Any) = Some("")
}
object Test {
def main(args: Array[String]): Unit = {
"" match {
case Single(x) =>
(x, x)
}
}
}
% qscalac -optimize -Xprint:patmat test/files/run/t9003b.scala 2>&1 | grep --context=5 get
case <synthetic> val x1: Any = "";
case5(){
<synthetic> val o7: Some[String] = Single.unapply(x1);
if (o7.isEmpty.unary_!)
matchEnd4({
scala.Tuple2.apply[String, String](o7.get, o7.get);
()
})
else
case6()
};
% scalac-hash v2.11.4 -optimize -Xprint:patmat test/files/run/t9003b.scala 2>&1 | grep --context=5 get
case <synthetic> val x1: Any = "";
case5(){
<synthetic> val o7: Some[String] = Single.unapply(x1);
if (o7.isEmpty.unary_!)
matchEnd4({
scala.Tuple2.apply[String, String](o7.get, o7.get);
()
})
else
case6()
};
```
|
| |\ \ \ \ \ \ \ \
| | | | | | | | |
| | | | | | | | | |
SI-9018 Fix regression: cycle in LUBs
|
| | | |_|/ / / / /
| |/| | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | | |
Regressed in 4412a92d, which admirably sought to impose some structure
on the domain of depths, but failed to preserve an imporatnt part of
said structure.
When calculating LUBs and GLBs, the recursion depth is limited by
propagating a decreasing depth parameter. Its initial value is
the recursion limit, and is calcluated from the maximum depth of
the types fed into the calculation.
Here are a few examples that give a flavour of this calculation:
```
scala> class M[A]
defined class M
scala> class N extends M[M[M[M[A]]]]
<console>:34: error: not found: type A
class N extends M[M[M[M[A]]]]
^
scala> class N extends M[M[M[M[Int]]]]
defined class N
scala> lubDepth(typeOf[N] :: Nil)
res5: scala.reflect.internal.Depth = Depth(4)
scala> type T = M[Int] with M[M[Int]]
defined type alias T
scala> lubDepth(typeOf[T] :: Nil)
res7: scala.reflect.internal.Depth = Depth(3)
```
One parts of the LUB calculation, `lub0`, truncates the lub to
`Any` when the depth dives below zero.
Before 4412a92d:
------------------
value decr incr
------------------
-3 -3 -2 (= AnyDepth)
-2 -3 -1
-1 -2 0
0 -1 1
1 0 2
...
After 4412a92d:
-----------------------
value decr incr
-----------------------
-MaxInt -MaxInt -MaxInt (= AnyDepth)
0 -MaxInt 1
1 0 2
...
The crucial difference that triggered the regression is that
decrementing a depth of zero now goes to the sentinel value,
`AnyDepth`, rather than to `-1`.
This commit modifies `Depth` to allow it to represent any negative
depth. It also switches the sentinel value for `AnyDepth`. Even
though I don't believe it is needed, I have also allowed for
`Depth.Zero.decr.decr.decr == Depth.AnyVal`, which was historically
the case in 2.10.4.
To better understand what was happening, I added tracing to the
calculation and diffed the before and after:
https://gist.github.com/retronym/ec59608eecc52bb497fa
Notice that when `elimSub(ts, depth = 0)` recursively calls `lub`,
it does so with the variant that caluculates the allowable depth
from the shape of the given types. We can then infinitely recurse.
Before 4412a92d:
```
|-- elimSub(depth = 0, ts = List(Comparable[_ >: TestObject.E.Value with String <: Comparable[_ >: TestObject.E.Valu
| |-- lub(depth = -1, ts = List(TestObject.E.Value with String, TestObject.C))
| | |-- lub0(depth = -1, ts0 = List(TestObject.E.Value with String, TestObject.C))
| | | |-- elimSub(depth = -1, ts = List(TestObject.E.Value with String, TestObject.C))
| | | |== List(TestObject.E.Value with String, TestObject.C)
| | | |-- Truncating LUB to
| | | |== Any
| | |== Any
| |== Any
|== List(Comparable[_ >: TestObject.E.Value with String <: Comparable[_ >: TestObject.E.Value with String <: java.io
|-- lub(depth = 0, ts = List(java.lang.type, java.lang.type))
| |-- lub0(depth = 0, ts0 = List(java.lang.type, java.lang.type))
| | |-- elimSub(depth = 0, ts = List(java.lang.type, java.lang.type))
| | |== List(java.lang.type)
| |== java.lang.type
|== java.lang.type
|-- elimSub(depth = 0, ts = List(Object, Object))
|== List(Object)
|-- elimSub(depth = 0, ts = List(Any, Any))
|== List(Any)
```
After 4412a92d:
```
|-- elimSub(depth = 0, ts = List(Comparable[_ >: TestObject.E.Value with String <: Comparable[_ >: TestObject.E.Valu
| |-- lub(depth = _, ts = List(TestObject.E.Value with String, TestObject.C))
| | |-- lub(depth = 3, ts = List(TestObject.E.Value with String, TestObject.C))
| | | |-- lub0(depth = 3, ts0 = List(TestObject.E.Value with String, TestObject.C))
| | | | |-- elimSub(depth = 3, ts = List(TestObject.E.Value with String, TestObject.C))
| | | | |== List(TestObject.E.Value with String, TestObject.C)
| | | | |-- lub1(depth = 3, ts = List(TestObject.E.Value with String, TestObject.C))
| | | | | |-- elimSub(depth = 3, ts = List(scala.math.Ordered[TestObject.E.Value], scala.math.Orde
| | | | | |== List(scala.math.Ordered[TestObject.E.Value], scala.math.Ordered[TestObject.C])
```
|
| |/ / / / / / / |
|
| | |_|/ / / /
|/| | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
When unpickling a class, we create stub symbols for references
to classes absent from the current classpath. If these references
only appear in method signatures that aren't called, we can
proceed with compilation. This is in line with javac.
We're getting better at this, but there are still some gaps.
This bug is about the behaviour when a package is completely
missing, rather than just a single class within that package.
To make this work we have to add two special cases to the unpickler:
- When unpickling a `ThisType`, convert a `StubTermSymbol` into
a `StubTypeSymbol`. We hit this when unpickling
`ThisType(missingPackage)`.
- When unpickling a reference to `<owner>.name` where `<owner>`
is a stub symbol, don't call info on that owner, but rather
allow the enclosing code in `readSymbol` fall through to
create a stub for the member.
The test case was distilled from an a problem that a Spray user
encountered when Akka was missing from the classpath.
Two existing test cases have progressed, and the checkfiles are
accordingly updated.
|
| | | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
It passed in PR validation but failed in a later run.
There are some clever ideas bouncing around to make it
stable, but in the meantime I'll shunt it into disabled.
|
| |/ / / / /
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
References to Threads would be retained long after their termination if
reflection is used in them. This led to a steady, long memory leak in
applications using reflection in thread pools.
|
| |\ \ \ \ \
| | | | | |
| | | | | | |
SI-5938 Test for a FSC bug with mixin, duplicate static forwarders
|
| | | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
Under resident compilation, we were getting multiple copies of static
forwarders created for mixed in methods. It seems like the bug was
fixed as a by-product of #4040.
This commit adds a test to show this. I've confirmed that the test
fails appropriately with 2.11.4.
For future reference, before I figured out how to write the test
for this one (test/resident doesn't seem to let you run the code
after compiling it), I was using bash to test as follows:
(export V=2.11.x; (scalac-hash $V sandbox/t5938_1.scala; (for i in 1 2; do echo sandbox/t5938.scala; done; printf '\n') | scalac-hash $V -Xresident); stty echo; scala-hash $V X ; echo ':javap -public X' | scala-hash $V);
|
| |\ \ \ \ \ \
| | | | | | |
| | | | | | | |
SI-7596 Curtail overloaded symbols during unpickling
|
| | | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
In code like:
object O { val x = A; def x(a: Any) = ... }
object P extends O.x.A
The unpickler was using an overloaded symbol for `x` in the
parent type of `P`. This led to compilation failures under
separate compilation.
The code that leads to this is in `Unpicklers`:
def fromName(name: Name) = name.toTermName match {
case nme.ROOT => loadingMirror.RootClass
case nme.ROOTPKG => loadingMirror.RootPackage
case _ => adjust(owner.info.decl(name))
}
This commit filters the overloaded symbol based its stability
unpickling a singleton type. That seemed a slightly safer place
than in `fromName`.
|
| |\ \ \ \ \ \ \
| |_|_|_|_|_|/
|/| | | | | | |
SI-8597 Improved pattern unchecked warnings
|
| | | |_|_|_|/
| |/| | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
The spec says that `case _: List[Int]` should be always issue
an unchecked warning:
> Types which are not of one of the forms described above are
> also accepted as type patterns. However, such type patterns
> will be translated to their erasure (§3.7). The Scala compiler
> will issue an “unchecked” warning for these patterns to flag
> the possible loss of type-safety.
But the implementation goes a little further to omit warnings
based on the static type of the scrutinee. As a trivial example:
def foo(s: Seq[Int]) = s match { case _: List[Int] => }
need not issue this warning.
These discriminating unchecked warnings are domain of
`CheckabilityChecker`.
Let's deconstruct the reported bug:
def nowarn[T] = (null: Any) match { case _: Some[T] => }
We used to determine that if the first case matched, the scrutinee
type would be `Some[Any]` (`Some` is covariant). If this statically
matches `Some[T]` in a pattern context, we don't need to issue an
unchecked warning. But, our blanket use of `existentialAbstraction`
in `matchesPattern` loosened the pattern type to `Some[Any]`, and
the scrutinee type was deemed compatible.
I've added a new method, `scrutConformsToPatternType` which replaces
pattern type variables by wildcards, but leaves other abstract
types intact in the pattern type. We have to use this inside
`CheckabilityChecker` only. If we were to make `matchesPattern`
stricter in the same way, tests like `pos/t2486.scala` would fail.
I have introduced a new symbol test to (try to) identify pattern
type variables introduced by `typedBind`. Its not pretty, and it
might be cleaner to reserve a new flag for these.
I've also included a test variation exercising with nested matches.
The pattern type of the inner case can't, syntactically, refer to the
pattern type variable of the enclosing case. If it could, we would
have to be more selective in our wildcarding in `ptMatchesPatternType`
by restricting ourselves to type variables associated with the closest
enclosing `CaseDef`.
As some further validation of the correctness of this patch,
four stray warnings have been teased out of
neg/unchecked-abstract.scala
I also had to changes `typeArgsInTopLevelType` to extract the type
arguments of `Array[T]` if `T` is an abstract type. This avoids the
"Checkability checker says 'Uncheckable', but uncheckable type
cannot be found" warning and consequent overly lenient analysis.
Without this change, the warning was suppressed for:
def warnArray[T] = (null: Any) match { case _: Array[T] => }
|
