| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| | |
|
| |
|
|
|
|
|
|
| |
Fixed the fact that reflective calls had logical and arithmetic right
shift swapped. Modified test case to use a value that doesn't have
identical output for both shifts. Grumbled to self that test cases which
fail to test are markedly worse than no tests at all.
|
| |
|
|
|
|
| |
Removed unused Any_asInstanceOfErased and Any_isInstanceOfErased
variations on is/as.
|
| |
|
|
|
| |
Hopefully fix for inadvertant stability failure introduced in r18118.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Initial checkin of a code generation DSL which I felt like I had to
undertake before I could begin the pattern matcher in good conscience.
I'm pretty excited about the impact this could have
on the understandability of the codebase - for me anyway.
// actual code in this checkin
DEF(method) === {
(This(clazz) EQREF that) OR (that MATCH(
(CASE(pat) IF guard) ==> TRUE ,
DEFAULT ==> FALSE
))
}
|
| |
|
|
|
|
| |
New jline jar from mapreduce's generified jline, including 32-bit and
64-bit versions of jline.dll so windows users can be happier.
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
| |
Reverted attempts at inhibiting empty .scala_dependencies
|
| |
|
|
|
|
|
| |
The dependecy file is read only when -make is not 'all'. Added a more
refined builder that tracks dependency at the member level, and moved
the build managers to package interactive.
|
| |
|
|
|
| |
named arguments only at top level in () parameters.
|
| | |
|
| |
|
|
|
|
|
| |
Separated default source code encoding (the one specified by -encoding)
from default system encoding, because otherwise you can't compile scala
with the defaults on OSX.
|
| |
|
|
|
|
| |
Generalizations made in the course of evaluating pattern matcher
rewrite.
|
| |
|
|
|
|
| |
fixed variance bug; added smart brace insertion to parser. moved
interactive compiler interface along.
|
| |
|
|
|
|
| |
More work and documentation for GenericRanges, plus minor findbugs
noticed issues.
|
| |
|
|
|
|
|
|
|
|
|
| |
Working my way toward a consistent interface for scala.io.Source.
Now all the Source factory methods whose input is something Byte
based rather than Char based require an implicit Codec, which will be
Codec.default if no implicit is available. The idea is that if one uses
the scala APIs, your desired encoding can be expressed as an implicit
value and it will always be used; and that we (if so inclined) can
change the default in one place.
|
| |
|
|
|
|
| |
Moving empty .scala_dependencies check around looking for the spot which
won't cause trouble.
|
| |
|
|
|
|
|
| |
* dialogs use named arguments now, messages are of type Any (fixes
* #1922) components register listeners on demand (fixes #1595) Fix for
* #2026 minor additions and fixes
|
| |
|
|
|
|
|
|
| |
Some enhancements to Option to cover a variety of cases which have
come up repeatedly: Option factory which translates null to None
.orNull method on Options of nullable types filterMap implementation so
Option[T].filterMap => Option[U]
|
| |
|
|
|
|
|
| |
scala.io.Codec for encapsulating charset encoding/decoding info, and
then uses that class as the world's first default/implicit parameter:
see BufferedSource.fromInputStream.
|
| |
|
|
|
|
|
|
|
|
| |
Made some minor adjustments to Stream, having deprecated methods
actually use the implementation they recommend to everyone else, except
in those cases where doing so caused scalac to become unbuildable (e.g.
concat) or the suggested alternative doesn't actually exist (e.g.
fill(elem)). Also dropped the "drop" override as it appeared identical
to the inherited implementation.
|
| | |
|
| |
|
|
|
| |
Thread-based receive uses ManagerBlocker interface of new ForkJoinPool.
|
| |
|
|
|
| |
Ordering extend Comparator.
|
| | |
|
| |
|
|
|
|
| |
Removed what amounted to a no-operation in the synthetic case class
equals method, and removed some dead code at the source level as well.
|
| |
|
|
|
|
|
|
| |
Attempted to resolve the improbable mess surrounding implicit
conversions from Unit. Modified test case which relied on the supposedly
verboten behavior; verbotenized () => AnyRef; added new test case which
fails if verboten behavior should ever return.
|
| |
|
|
|
|
|
| |
and the various Int types all use different hashCode algorithms. This
means there is very broken behavior afoot among the types. I see no
option but to make numeric equals significantly less tricky.
|
| |
|
|
|
| |
Created DelayedLazyVal for your def/val hybrid needs.
|
| | |
|
| |
|
|
|
|
| |
improvements to names / defaults (implicits, type of defaults, #2064,
...)
|
| |
|
|
|
| |
Added jline as optional dependency to scala-compiler pom
|
| |
|
|
|
|
| |
Removed some useless type arguments which survived an earlier iteration,
and resolved an implicit ambiguity that snuck into SpecializeTypes.
|
| |
|
|
|
| |
class Foo[@specialized("Int, Double") T] ..
|
| |
|
|
|
|
|
|
|
| |
Created scala.reflect.Invocation, which offers an easy syntax for
reflective method calls, like: "abcdefgh" o 'endsWith("gh") // returns
Any
val res: Boolean = "abc" oo 'endsWith("z") // casts to expected type
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
| |
Small fixes in code generation for interface calls.
|
| |
|
|
|
|
| |
Fixed symbol loaders when called at a very late phase (for example, when
inner classes are lifted).
|
| |
|
|
|
|
| |
(1) some changes to interactive compiler interface. 2) added
(symbol.hasTypeAt 3) Added flatten/transpose/unzip to TraversableClass
|
| |
|
|
|
|
|
| |
Removed some useless files which declare themselves non-existent for the
dotnet target, clearly without realizing they don't exist for any other
target either.
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
| |
Adds @experimental annotation to lower the activation energy wall
between me and all my good ideas. Then took advantage of same to add
the experimental filterMap to Traversable. (I consider its goodness
undeniable, but its name at least is experimental.)
|
| |
|
|
|
|
| |
Newly generated Tuple/Product/Function classes from new genprod, plus a
couple minor tweaks to genprod to keep tests happy.
|
