| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
| |
Partest uses test-interface API to invoke ScalaCheck. This obviates
ad hoc output checking for result status.
The context class loader is set to a loader that the scaladoc scalacheck
tests can use.
|
|\
| |
| | |
SI-7291: Don't throw exceptions while encountering diverging expansion.
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Since we don't throw exceptions for normal errors it was a bit odd
that we don't do that for DivergingImplicit.
As SI-7291 shows, the logic behind catching/throwing exception
was broken for divergence. Instead of patching it, I rewrote
the mechanism so that we now another SearchFailure type related
to diverging expansion, similar to ambiguous implicit scenario.
The logic to prevent diverging expansion from stopping the search
had to be slightly adapted but works as usual.
The upside is that we don't have to catch diverging implicit
for example in the presentation compiler which was again showing
that something was utterly broken with the exception approach.
|
| |
| |
| |
| |
| |
| | |
Sick of seeing Console printlns during partest runs.
You should not print anything to Console.{out,err} if
it's ever going to happen outside developerland.
|
|\ \
| | |
| | | |
-Yshow-member-pos, print the positions of members.
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Here for instance is a command line which leverages the
output of this option to print the method bodies of all
methods called 'transformInfo' found under src. Given
1500 source files it accomplishes this in four seconds,
thanks to -Ystop-after:parser.
% scalac -Yshow-member-pos sed -Ystop-after:parser \
$(find src/compiler -name '*.scala') | \
grep transformInfo | sed 's/ # .*//;' | \
while read line; do echo "// $line" && gsed -n $line && echo; done
Or more simply, the start/end lines of each member of Random:
% scalac -Yshow-member-pos "" ./src/library/scala/util/Random.scala
./src/library/scala/util/Random.scala
20,134 class Random
33 def nextBoolean
38 def nextBytes
43 def nextDouble
48 def nextFloat
54 def nextGaussian
59 def nextInt
65 def nextInt
70 def nextLong
81,89 def nextString
82,86 def safeChar
83 val surrogateStart
84 val res
94,98 def nextPrintableChar
[snip]
It makes me sad I'm always in the position of having to hack
the compiler to do this sort of thing. All we need is something
like -Yinsert-phase:Foo where Foo is a class implementing a
(Phase, Tree) => Tree method, and the compiler runs all the
unit.bodies through it after each phase, and then one could
easily accomplish this in the privacy of one's own compiler.
|
|\ \ \
| | | |
| | | | |
SI-7403 Stream extends Serializable
|
| | |/
| |/|
| | |
| | |
| | | |
Additionally, add @deprecatedInheritance to warn creators of
custom subclasses that the class will be sealed in the future.
|
|\ \ \
| |/ /
|/| | |
SI-4365 nondeterministic failure in asSeenFrom
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Under some order-dependent conditions (if source files arrive
in one order it happens, in the other order it does not) more
than one set of type parameters are created for a given class.
Previously this would lead to a crash in asSeenFrom when a type
parameter had to be matched up with a type application.
Now when that situation arises I compare them by name and log
a dev warning if it hits. This does not risk anything undesirable
happening because the wayward type parameter's owner is always
the right class; it's only the class type parameters which don't
include the wayward one. Since in a given type parameter list
names are unique, we have enough information to salvage the
search.
|
|\ \ \
| | | |
| | | |
| | | |
| | | | |
Conflicts:
src/compiler/scala/tools/nsc/typechecker/Contexts.scala
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
While we are all aware of the issues around Serialization,
I think in this case it is perfectly sound and safe to make
List serializable:
- List is not an interface, it is the base type of an ADT.
Common behavior of its members should be reflected in the base type.
- List is sealed, there is no chance of an user providing a new
non-serializable subtype of List.
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
It's a couple orders of magnitude out of whack when a test
demands a gigabyte of memory to pass. We might need to start
collecting per-test stats to avoid this kind of thing in the
future. It's a huge waste of time.
|
|/ / /
| | |
| | |
| | |
| | |
| | | |
Some unused private code, unused imports, and points where
an extra pair of parentheses is necessary for scalac to have
confidence in our intentions.
|
|\ \ \
| | | |
| | | | |
Simplify type bounds.
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
I started out looking to limit the noise from empty type
bounds, i.e. the endless repetition of
class A[T >: _root_.scala.Nothing <: _root_.scala.Any]
This led me to be reminded of all the unnecessary and
in fact damaging overreaches which are performed during parsing.
Why should a type parameter for which no bounds are
specified be immediately encoded with this giant tree:
TypeBounds(
Select(Select(Ident(nme.ROOTPKG), tpnme.scala_), tpnme.Nothing),
Select(Select(Ident(nme.ROOTPKG), tpnme.scala_), tpnme.Any)
)
...which must then be manually recognized as empty type bounds?
Truly, this is madness.
- It deftly eliminates the possibility of recognizing
whether the user wrote "class A[T]" or "class A[T >: Nothing]"
or "class A[T <: Any]" or specified both bounds. The fact
that these work out the same internally does not imply the
information should be exterminated even before parsing completes.
- It burdens everyone who must recognize type bounds trees,
such as this author
- It is far less efficient than the obvious encoding
- It offers literally no advantage whatsoever
Encode empty type bounds as
TypeBounds(EmptyTree, EmptyTree)
What could be simpler.
|
| |/ /
|/| |
| | |
| | |
| | |
| | | |
run/t4023 can fail (and has already) because the order of elements
in the reflection API is not specified and can differ between platforms
(e. g. HotSpot and Avian) or even versions (Java 7 vs. Java 8).
|
|\ \ \
| | | |
| | | | |
SI-7376 Scaladoc warns when discarding local doc comments with API tags
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
The unmoored DocComment is created more eagerly so that its position
is correct despite subsequent line comments. (Previously, skipComment
would advance docPos.)
It looks like the error caret is still off by one when a doc comment
shows up in the middle of an operator, and who doesn't scaladoc the
interior of expressions?
Another bug fixed by Paul's refactor is that additional comments
between the doc and the entity no longer breaks the scaladoc.
Test added.
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Double-star doc comments in non-dockable positions at the end of a block
will emit a warning only if API tags like @author are present, or under
-Xlint.
A real comment parser is applied early to probe for tags, to minimize
ad hoc testing or duplication, but warnings are suppressed. Residual
ad hockiness lies in precisely which tags to warn on. Ad hoc or ad doc.
This fix is a stop gap; a richer solution would also report about other
doc locations that won't be processed.
|
|\ \ \ \
| | | | |
| | | | | |
SI-7080 improve boundary value checking for BitSet
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
When BitSet accepts a very large integer such as Int.MaxValue,
integer overflow possibly occurs in the calculation of boundary
value "nwords * WordLength". This faulty boundary condition
causes empty-iterator problem like following:
scala> import collection.mutable.BitSet
import collection.mutable.BitSet
scala> val x = BitSet(Int.MaxValue)
x: scala.collection.mutable.BitSet = BitSet()
scala> x.iterator
res0: Iterator[Int] = empty iterator
|
| |_|_|/
|/| | |
| | | |
| | | | |
Fail those monster methods rather than generating bad bytecode.
|
|\ \ \ \
| | | | |
| | | | | |
SI-7337 Error out on missing -d directory.
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
This check was removed without comment in 3a30af154, the addition
of JSR-223 support for the interpreter.
After this commit, I manually tested that JSR-223 support works.
scala> import javax.script._, collection.JavaConverters._; val manager = new ScriptEngineManager; manager.getEngineByName("scala").eval("List(1)")
import javax.script._
import collection.JavaConverters._
manager: javax.script.ScriptEngineManager = javax.script.ScriptEngineManager@4418f61b
res1: Object = List(1)
3a30af154 did not include a test case, so I don't know whether I've
broken some other aspect of it. I tried the above as a `run` test,
but hit two problems, one of them seemingly our fault, and the other
a MacOS JDK wrinkle.
1. scala.reflect.internal.MissingRequirementError: object scala.runtime in compiler mirror not found.
2. java.lang.UnsatisfiedLinkError: no AppleScriptEngine in java.library.path
I can't find my way to fix these, so JSR-223 remains untested. I don't
think that commit was really up to standard; it could handle additional
review, documentation, and testing. It might even be modularized so as
not to pollute the REPL itself.
|
|\ \ \ \ \
| |_|_|_|/
|/| | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
merge/v2.10.1-245-g5147bb2-to-master
Conflicts:
src/compiler/scala/tools/nsc/transform/SpecializeTypes.scala
src/compiler/scala/tools/nsc/typechecker/Typers.scala
|
| |\ \ \ \
| | | | | |
| | | | | | |
Absolute path in error message.
|
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
As soon as you have a directory called "language" lying around,
you will appreciate why the advice given regarding SIP-18
should be "import scala.language..." not "import language..."
|
| |\ \ \ \ \
| | | | | | |
| | | | | | | |
SI-7388 Be more robust against cycles in error symbol creation.
|
| | |/ / / /
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
`Symbol#toString` was triggering `CyclicReferenceError` (specifically,
`accurateKindString` which calls `owner.primaryConstructor`.)
The `toString` output is used when creating an error symbol to
assign to the tree after an error (in this case, a non-existent
access qualifier.)
This commit catches the error, and falls back to just using the
symbol's name.
|
| |\ \ \ \ \
| | | | | | |
| | | | | | | |
SI-7377 Fix retypechecking of patterns on case companion alias
|
| | |/ / / /
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
Some ancient code in Typers switches from PATTERNmode to
EXPRmode when encountering `stableF(...)`. It just
typechecks `stableF` and discards the arguments.
To the best of Martin's recollection, this has something
to do with the need to typecheck patterns rather late in
the compiler, after `a.b` had been translated to `a.b()`
in `Uncurry`. I'm not able to motivate this with tests
using `-Xoldpatmat`; was there ever an even older pattern
matcher that ran *after* uncurry?
What changed in 2.10.1 to expose this wrinkle? dfbaaa17
fixed `TypeTree.copyAttrs` to copy the original tree.
During the descent of `ResetAttrs`, sub-trees are duplicated
before begin further transformed. Duplicating the `Match`
in 2.10.0 would forget that the original tree of:
pat = (a: Int)Foo(_)
`----------`
`- TypeTree((a: Int)Foo), with original Select(..., "FooAlias")
The retypechecking would operate on the `MethodType`, rather
than the `Select`, which was not considered a stable
application.
For 2.10.x, I've just tightened up the condition to only
hit this if `args` is empty. I'm almost certain that the code
can be removed altogether, and I'll do that when this is merged
to master.
|
| |\ \ \ \ \
| | | | | | |
| | | | | | | |
SI-7319 Avoid unflushed error/warning buffers in startContext
|
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
Contexts share error/warning buffers with their
children, and this also applies ot the shared
`startContext`.
That context flushes the buffers in `startContext`
in `resetContexts`.
It also removes `typerReportAnyContextErrors`, which
appears to be an elaborate no-op. It is only ever passed
a context `c` which is a direct child of `this.context`.
So taking a buffered error out of `c` and reissuing it
into `this.context` merely re-inserts into into the same
error buffer. Consrast this with `silent`, which uses
a child context with a fresh error buffer.
SI-7319 Flush error buffer in typerReportAnyContextErrors.
After this change, we no longer rely on the backstop in resetContexts
introduced in the previous commit.
|
| |\ \ \ \ \ \
| | |_|/ / / /
| |/| | | | | |
SI-7329 duplicate default getters for specialized parameters.
|
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
The default getter is generated with @specialized annotation if the
type parameter corresponding to the type of the parameter is
specialized. Consequently specialize pass tries to generate overloads.
Rather than pruning overloads to exclude duplicates, let's notice
that default getter specialization is not needed at all:
- The dynamic scope of default getter doesn't include specialized
method or class constructor.
- generic default getter is called even when calling specialized
method:
object V {
@specialized def foo[@specialized B](b: B = (??? : B)) = {}
foo[Int]()
}
gives:
invokevirtual Method V$.foo$default$1:()Ljava/lang/Object;
invokestatic (unboxToInt)
invokevirtual Method V$.foo$mIc$sp:(I)V
|
|\ \ \ \ \ \ \
| | | | | | | |
| | | | | | | | |
Remove scaladoc deprecated option.
|
| | |_|_|_|_|/
| |/| | | | |
| | | | | | |
| | | | | | | |
Remove -external-urls that had been previously deprecated in 2.10.
|
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
This commit lets partest locate tools.jar the way REPL does, with
the addition that java.home.parent is also tried.
The partest script will use JAVAC_CMD if set, or else JAVA_HOME, and
will try the sibling of JAVACMD if set (on the theory that if you specify
java, you are avoiding the path lookup and javac may also be in
that special place), or else query the path for javac.
The use cases are: no env vars, look around java.home; JDK or JAVA_HOME is
set; JAVACMD is set; and finally tools.jar can live in jre/lib/ext and
the fallback deep search will find it.
These cases have been tried on cygwin with Java installed under
s"Program${space}Files", which is usually the most brittle environment.
That means tested with bash.
The windows partest.bat has not been upgraded or side-graded.
|
|\ \ \ \ \ \ \
| | | | | | | |
| | | | | | | | |
Reifier -> AST Node test.
|
| |/ / / / / /
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
It's a Node-by-Node tour of the reifier's abilities and
occasional foibles.
That is one spectacularly attractive checkfile.
|
|\ \ \ \ \ \ \
| | | | | | | |
| | | | | | | | |
SI-7312 @deprecatedInheritance now ignores same-file subclasses
|
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | | |
Don't warn about the specialized subclass.
Fixed in the previous commit for a similar issue SI-7312.
|
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | | |
This allows us to deprecate external inheritances as a prelude
to sealing a class, without enduring the warnings ourselved in
interlude.
|
|\ \ \ \ \ \ \ \
| |_|/ / / / / /
|/| | / / / / /
| | |/ / / / /
| |/| | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
merge/v2.10.1-235-g4525e92-to-master
Conflicts:
bincompat-backward.whitelist.conf
bincompat-forward.whitelist.conf
src/compiler/scala/tools/nsc/transform/SpecializeTypes.scala
src/compiler/scala/tools/nsc/typechecker/Typers.scala
src/reflect/scala/reflect/internal/Types.scala
|
| | |_|/ / /
| |/| | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
Specialize assigns SpecialOverride info to a specialized method
even when there is a further specialization that should be forwarded
to.
|
| |\ \ \ \ \
| | | | | | |
| | | | | | | |
Revert "SI-6387 Clones accessor before name expansion"
|
| | | |/ / /
| | |/| | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
This reverts commit 4e10b2c833fa846c68b81e94a08d867e7de656aa.
Add 6387 test to pending and 7341 to up-to-date.
|
| |\ \ \ \ \
| | | | | | |
| | | | | | | |
SI-6386 typed existential type tree's original now have tpe set
|
| | |/ / / /
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
Tree reification fails for ExistentialTypeTree. The reason is that the
tree passed for reification (see reifyTree at GetTrees.scala) must have
not null tpe (see reifyBoundType at GenTrees.scala), which is not true
in the case of ExistentialTypeTree.
Why is it so? The tree passed to reifyTree was obtained in the reshape
phase of reificationusing using original TypeTrees that reporesent pre-
typer representation of a type. The problem is that original's tpe for
ExistentialTypeTree is not set.
So the solution to the issue is to create ExistentialTypeTree's original
in a such way that is has actual tpe set.
|