| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Post-erasure of value classs in method signatures to the underlying
type wreaks havoc when the erased signature overlaps with the
generic signature from an overriden method. There just isn't room
for both. But we *really* need both; callers to the interface method
will be passing boxed values that the bridge needs to unbox and
pass to the specific method that accepts unboxed values.
This most commonly turns up with value classes that erase to
Object that are used as the parameter or the return type of
an anonymous function.
This was thought to have been intractable, unless we chose
a different name for the unboxed, specific method in the
subclass. But that sounds like a big task that would require
call-site rewriting, ala specialization.
But there is an important special case in which we don't need
to rewrite call sites. If the class defining the method is
anonymous, there is actually no need for the unboxed method;
it will *only* ever be called via the generic method.
I came to this realisation when looking at how Java 8 lambdas
are handled. I was expecting bridge methods, but found none.
The lambda body is placed directly in a method exactly matching
the generic signature.
This commit detects the clash between bridge and target,
and recovers for anonymous classes by mangling the name
of the target method's symbol. This is used as the bytecode
name. The generic bridge forward to that, as before, with
the requisite box/unbox operations.
|
| |\
| |
| | |
SI-8092 More verify for f-interpolator
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Attempt to verify the nooks and crannies of the format string.
Allows all syntax in the javadoc, including arg indexes. If the
specifier after an arg has an index that doesn't refer to the arg,
a warning is issued and the missing `%s` is prepended (just as
for a part with a leading `%n`).
Other enhancements include detecting that a `Formattable` wasn't
supplied to `%#s`.
Error messages attempt to be pithy but descriptive.
|
| |\ \
| | |
| | | |
SI-8131 fixes residual race condition in runtime reflection
|
| | | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Apparently some completers can call setInfo while they’re not yet done,
which resets the LOCKED flag, and makes anything that uses LOCKED to
track completion unreliable. Unfortunately, that’s exactly the mechanism
that was used by runtime reflection to elide locking for symbols that are
known to be initialized.
This commit fixes the problematic lock elision strategy by introducing
an explicit communication channel between SynchronizedSymbol’s and their
completers. Now instead of trying hard to infer whether it’s already
initialized or not, every symbol gets a volatile field that can be queried
to provide necessary information.
|
| |\ \ \
| | | |
| | | | |
SI-4997 deprecate StringLike.linesIterator for StringLike.lines
|
| | | | |
| | | |
| | | |
| | | | |
Deprecated. lines is by far more consistent with the rest of the naming in the library.
|
| |/ / /
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Regressed in SI-7015 / 1b6661b8.
We do need to "unbox" the null (ie, drop a stack from and load
a null) in general. The only time we can avoid this is if the
tree we are adapting is a `Constant(Literal(null))`.
I've added a test for both backends. Only GenICode exhibited
the problem.
|
| |\ \ \
| | | |
| | | | |
SI-7700 @unspecialized, Part Deux: Now Working.
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
This annotation was introduced to allow us to mark methods within
a specialized trait as immune from specialization. In particular,
this is desirable for `Function1.{andThen, compose}`.
However, it seems we need to check for this in two places in the
specialization code base. The feature is now backed with a test.
|
| |\ \ \ \
| | | | |
| | | | | |
SI-8143 Regressions with override checks, private members
|
| | | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
These regressed in e609f1f20b, which excluded all private methods from
overriding checks. We should only exclude private[this] members on the
low end of a pair, as was done before that commit, and, we must also
exclude private members on the high side.
Why? Warning: reverse engineered intuition follows.
We need to report an error when if a private method in a subclass
has matches a less-private method in the super class and report an
error, lest the user be fooled into thinking it might be invoked
virtually. On the other hand, adding a private method to a super
class shouldn't invalidate the choice names of public members in
its superclasses.
I've removed the test case added by that commit and will lodge a
reworked version of it that Paul provided as a new issue. That shows
a bug with qualified private + inheritance.
In addition, the expectation of `neg/accesses.check` is reverted
to its 2.10.3 version, which I believe is correct. When it was
changed in e609f1f20b it sprouted a variation, `neg/accesses-2`,
which has now changed behaviour. The intent of that test will
be captured in the aforementioned issue covering qualified private
inheritance.
|
| | | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
The class file name of an inner class is based on the flattened
name of its owner chain.
But, if this is going to be unreasonably long, it is shortened
with the help of a MD5 hash.
However, after this shortening takes place, we sneakily add one
more character (the infamous '$') to the name when it is used
for the module class. It is thus possible to exceed the limit
by one.
The enclosed test failed on Mac with "filename too long" because
of this. I have also tested for trait implementatation classes,
but these seem to be suffixed with "$class" before the name
compactification runs, so they weren't actually a problem.
This change is binary incompatible as separately compiled
defintions and usages of named, inner classes need to agree
on this setting. Most typically, however, these long names
crop up for inner anonymous classes / functions, which are
not prone to the binary incompatiblity, assuming that their
creation hasn't be inlined to a separately compiled client.
|
| |\ \ \ \ \
| |_|_|_|/
|/| | | | |
ListMap.tail is returning wrong result
|
| | | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
Reverted the commit that introduced the bug, and modified HashMap to no
longer assume that tail is O(1).
Review by @Ichoran, @soc
|
| |\ \ \ \ \
| | | | | |
| | | | | | |
SI-6879 improves Context.freshName
|
| | | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
Instead of per-compilation unit unique counters, the freshName API now
uses a per-Global counter. Fresh names now also contain dollars to exclude
clashes with supported user-defined names (the ones without dollar signs).
This doesn’t fix the bug, because per-Global counters get created anew
every time a new Global is instantiated, and that provides some potential
for name clashes even for def macros, but at least it completely excludes
clashes in typical situations.
|
| | | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
In 2.11 we've changed TupleClass, ProductClass and FunctionClass
endpoints to be exposed as (Int => Symbol) functions that never throw
exceptions but rather return NoSymbol instead of previous error-prone
indexed access on array that could explode.
While simplifying one use case (indexed access) it complicated
ability to check if symbol at hand is in fact a tuple, product or
function:
(1 to 22).map(TupleClass).toList.contains(symbol)
To cover this extra use case we add a seq method to the variable arity
class definitions that exposes a corresponding sequence of class symbols:
TupleClass.seq.contains(symbol)
|
| |\ \ \ \ \ \
| |/ / / / /
|/| | | | | |
Improve naming of ReadStdin
|
| | | |_|/ /
| |/| | |
| | | | |
| | | | |
| | | | | |
We don't need to say everything twice, methods are already prefixed
with “read”.
|
| |\ \ \ \ \
| | | | | |
| | | | | | |
deprecates resetAllAttrs and resetLocalAttrs in favor of the new API
|
| | | |_|/ /
| |/| | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
We now have c.untypecheck, which is supposed to be a counterpart of c.typecheck
in the sense that it goes back from typed trees to untyped ones:
http://stackoverflow.com/questions/20936509/scala-macros-what-is-the-difference-between-typed-aka-typechecked-an-untyped.
Let’s hope that c.untypecheck will soon be able to solve our problems
with partially/incorrectly attributed trees emitted by macros:
https://groups.google.com/forum/#!topic/scala-internals/TtCTPlj_qcQ.
|
| |\ \ \ \ \
| | | | | |
| | | | | | |
SI-8151 Remove -Yself-in-annots and associated implementation
|
| | |/ / / /
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
This experimental option typechecked arguments of annotations
with an injected value in scope named `self`:
@Foo(self.foo < 1)
This has been slated for removal [1] for some time.
This commit removes it in one fell swoop, without any attempt
at source compatibility with code that constructs or pattern
matches on AnnotatedType.
[1] https://groups.google.com/d/msg/scala-internals/VdZ5UJwQFGI/C6tZ493Yxx4J
|
| |\ \ \ \ \
| | | | | |
| | | | | | |
Fix regression with package objects, overloading
|
| | |/ / / /
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
Regressed in f5c336d56, a refactoring of `typedIdent`. In that
commit, an (ostensibly) accidental change arrived, equivalent to:
- val pre1 = if (qual == EmptyTree) NoPrefix else if (sym.isTopLevel) sym.owner.thisType else qual.tpe
+ val pre1 = if (sym.isTopLevel) sym.owner.thisType else if (qual == EmptyTree) NoPrefix else qual.tpe
Here, `qual` is a tree returned in the successful result of
`Context#lookup`.
This change itself looks innocuous (top level symbols can be prefixed
with a qualifier or not, right?), but it exposed us to a bug in
`makeAccessible`. It is responsible for rewriting, e.g,
`scala.List` to `scala.package.List`. It has a few cases, but one
of them relies relies on typechecking `Ident(nme.PACKAGE)`, and
hoping that it will bind to the right place. That's fraught with
danger, and breaks in the enclosed tests.
This commit binds that Ident symbolically, and in the process
factors a tiny bit of code in common with `TreeGen`. (More work
is still needed here!)
In the next commit, I'm going to revert the change to `pre1`. That
would have also fixed the regression, albeit symptomatically.
|
| | | | | | |
|
| | |_|/ /
|/| | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Apparently, even though the compiler has safeguards against infinite type
printouts, having a depth counter, we didn’t account for the cases when
printouts are both self-referential and self-multiplying.
For one, SI-8158 provides an example of such a type, which is a structural
type that refers to itself twice in return types of its methods. At first,
printing such a type would go deeper and deeper, but then it will hit
the depth limit and start multiply indefinitely.
This commit fixes this particular problem by recognizing self-references
as this.type’s and printing them out as such. The subsequent commit will
introduce a more general facility.
|
| |\ \ \ \
| |_|/ /
|/| | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Check files updated: test/files/presentation/t8085*.check
Conflicts:
build.xml
src/compiler/scala/tools/nsc/ast/parser/Parsers.scala
src/compiler/scala/tools/nsc/symtab/classfile/ICodeReader.scala
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
We used to silently skip class files that would exceed the JVM's size limits.
While rare, this should still be an error.
While I was at it, also included the name of the offending method.
|
| |\ \ \ \
| | | | |
| | | | | |
SI-6253 HashSet should implement union
|
| | | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
Implements of HashSet.union that reuses the two trees as much as
possible when calculating the union of two sets. This leads to significant
performance improvements as well as to much better structural sharing.
There is a comprehensive correctness test for union since there was not a single test
for HashSet.union before. In addition, there are some tests of the desirable
properties of the new implementation (structural sharing and efficiency
regarding calls of key.hashCode).
The other operations diff and intersect, which are conceptually very
similar to union, are also implemented along with comprehensive test cases
for both correctness and structural sharing.
Note that while it appears that there is some code duplication between the
three methods, they are sufficiently different that it is not possible
to merge them into one without sacrificing performance.
|
| |/ / / /
| | | |
| | | |
| | | | |
Added extra ()'s to get rid of deprecation warning for inferring unit arg.
|
| |\ \ \ \
| | | | |
| | | | | |
SI-6196 - Set should implement filter
|
| | | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
This is the exact same algorithm as in SI-6196, with only slight
differences due to the two type arguments of HashMap.
Filter is tested by the new comparative collection test by @Ichoran, but
there is nevertheless a small correctness test in t6200 in addition to
tests of the new desirable behavior.
|
| | | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
Implements a version of filter and filterNot that reuses as much as
possible from the existing tree instead of building an entirely new one
like the builder-based filter does. This results in significant performance
improvements on average.
Adds a test of basic correctness of filter and filterNot as well as of the
desirable properties of the new filter implementation.
This is a collaboration between me and @Ichoran
|
| |\ \ \ \ \
| | | | | |
| | | | | | |
Quasi-comprehensive BigDecimal soundness/correctness fix.
|
| | | |_|_|/
| |/| | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
This fixes issues SI-6153, SI-6173, SI-6456, SI-6699, and SI-8116, along with a number of other similar possible issues.
Relevant changes include
* Changes to avoid heap explosion when working with BigInt
- to isWhole
- to hashCode
- to equals
- to BigInt's equals
* Changes to enable equality matching hashCode
- Only for sufficiently small BigInt
- For identical values with different precision
* Changes to isValidDouble
- Takes precision into account now
- New methods added to test whether even if the Double is not represented exactly, it's a representation of a certain type
- New companion methods added to allow intended expansion of Double (binary/decimal difference)
* Changes to constructor
- Null arguments are not allowed (these can throw NPEs later at awkward/unexpected times)
* New JUnit test to test all these things
* Fixed existing tests to expect new behavior
* Modified scaladocs to explain the issues
* Deprecated problematic methods
* Made application of MathContext more consistent (it is where you expect it and not where you don't)
These changes are coordinated, for the most part, hence the monolithic
commit.
|
| |\ \ \ \ \
| | | | | |
| | | | | | |
SI-8100 - prevent possible SOE during Stream#flatten.
|
| | | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
This commit changes stream flatten to avoid allocating a
stack frame for every stream item. Previously, flattening
a stream whose elements are mostly empty would result in
a StackOverflowException.
This commit also adds a test demonstrating the problem.
|
| |\ \ \ \ \ \
| | | | | | |
| | | | | | | |
Count lines by EOLs
|
| | | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
Source lines were counted by "line break chars", including FF.
Clients of `pos.line` seem to all expect the ordinary line num,
so that is what they get.
Unicode processing now precedes line ending processing.
|
| |\ \ \ \ \ \ \
| |_|_|_|/ / /
|/| | | | | | |
Deprecate automatic () insertion in argument lists
|
| | |/ / / / /
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
This promotes the () insertion warning from -Ywarn-adapted-args to a
deprecation warning. -Xfuture tunrs it into a compiler error.
Auto tupling remains unchanged for now.
The tests have been fixed the following way:
- Warnings caused by general sloppiness (Try(), Future(), ...) have been
fixed.
- Warnings which raise interesting questions (x == (), ...) received an
updated checkfile for now.
|
| |\ \ \ \ \ \
| | | | | | |
| | | | | | | |
Add tree-based code generation
|
| | | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
def printTypeParams is modified. Tests are updated.
|
| |\ \ \ \ \ \ \
| |_|_|_|/ / /
|/| | | | | | |
Implements specialized subsetOf for HashSet
|
| | | |_|/ / /
| |/| | | |
| | | | | |
| | | | | | |
Fixes SI-7326. This also adds a basic test for subsetOf that was missing before.
|
| |\ \ \ \ \ \
| | | | | | |
| | | | | | | |
reshuffles names for blackbox/whitebox contexts, changes bundle notation
|
| | | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
Adjusts bundle notation to read `class Bundle(val c: Context)` instead of
`class Bundle extends Macro`. This avoids calling compileLate in the
macro compiler and associated tooling problems.
|
