| Commit message (Collapse) | Author | Age | Files | Lines |
... | |
|
|
|
|
| |
Reverts a change as a tribute to the performance gods, no review.
|
|
|
|
|
|
|
|
|
| |
Created named subclasses not nested inside objects for various uniqified
Types. This has a large effect on memory consumption by avoiding an
unnecessary $outer field for each instance. (It also makes stack traces
look a lot nicer.) Statistics to be supplied eventually. Review by
odersky.
|
|
|
|
|
|
| |
Regression in Enumeration caused some nullary methods to pose as fields.
Closes #4570, no review.
|
| |
|
|
|
|
|
|
| |
Now that I've convinced myself getOrElseUpdate isn't a big killer
performancewise, simplified ZipArchive a bit. No review.
|
|
|
|
|
|
| |
Commented out a bug workaround which seems to be too costly to impose on
the general case. No review.
|
| |
|
|
|
|
|
|
| |
Created a PlainDirectory subclass of PlainFile so I can use polymorphism
to optimize. No review.
|
|
|
|
|
|
| |
Array[Byte] => Array[Char] => String => Array[Char] => Name. That's not
how Professor Optimize said to do it! No review.
|
|
|
|
|
| |
Eliminating more calls requiring filesystem access, no review.
|
|
|
|
|
|
| |
Wow, ZipEntry#getTime is expensive! So that's how I gave back all my
ZipArchive gains. No review.
|
|
|
|
|
| |
Uglified an unpickler method based on profiler input, no review.
|
|
|
|
|
|
|
|
|
| |
Makes MarkupParser.nextch lazy, only reaching out for the next char when
calling ch or eof. To make it possible, nextch now returns Unit instead
of Char. As it happens, that's how it is used almost everywhere.
Contributed by Daniel Sobral, no review.
|
|
|
|
|
|
|
|
|
|
| |
Moved scala.reflect.Chars to scala.reflect.internal.Chars, un-splitting
the scala.reflect package. This is necessary for running in an OSGi
container, such as Eclipse. A package may not be contributed by two
different bundles (in this case the library and the compiler). It may
be moved back when 'scala.reflect' exists only in one place. review by
extempore.
|
|
|
|
|
| |
Fix documentation of ## regarding null.##, thanks ymasory, no review.
|
|
|
|
|
|
|
|
|
|
|
|
| |
Removes SUnit (long deprecated!) from the standard library.
the relatively small number of partest tests in Scala's suite that
were still using SUnit now either just use regular asserts, or they
print stuff that partest checks with a .check file. Also fixed some bad
indentation, removed ancient useless-looking commented-out code, etc.
Contributed by Seth Tisue (way to go seth) no review.
|
|
|
|
|
|
|
| |
Apparent assumption that Literal(0) would be adapted to Literal(0.0f) as
necessary mercilessly invalidated. Fixed mkZero to account for all types
explicitly. Closes #4617, no review.
|
|
|
|
|
|
| |
Specialized keySets for immutable and sorted maps are supposed to be
views, not actual sets. Closes #4616, no review.
|
|
|
|
|
| |
No review.
|
| |
|
|
|
|
|
|
|
|
| |
Apparently some Windows tools are able to detect the fact that the .tgz
is really a .gz archive of one file, which originally did not have the
.tar suffix. That is pretty unusual, since even "gz -l" will list the
.tar suffix. In any event, this small change should make everyone happy.
|
| |
|
|
|
|
|
|
| |
Segregate deprecated members into a titled section. Contributed by
Daniel Sobral. review by malayeri
|
| |
|
|
|
|
|
|
| |
Error reporting when the generated code size exceeds JVM limits (65,535
bytes per method). Closed #4573. review by extempore.
|
|
|
|
|
|
|
|
| |
Eliminated an unnecessary second traversal of jars by reading classes
and packages in the same pass. And rewrote ZipArchive yet again. You'd
think I'd get tired of working on this file, but even now I don't think
we're quite there. No review.
|
|
|
|
|
|
| |
Deleting the empty files git is not too cool to remove but which somehow
evaded my steely gaze, no review.
|
|
|
|
|
|
| |
it wasn't buying anything. Clearly I won't be happy unless I never
encounter this class in the profiler again. No review.
|
|
|
|
|
|
|
| |
Renamed scala.reflect.common to scala.reflect.internal to better
emphasize that it is not API. (The brush was overly broad, and some
files now need to be rescued from being internal.) No review.
|
|
|
|
|
|
| |
And the remainder of the scala.reflect refactoring (think of it like a
"balloon payment") no review.
|
| |
|
| |
|
|
|
|
|
| |
Switched nsc.util to point to reflect.util, no review.
|
| |
|
|
|
|
|
| |
Some largely self-contained utility classes into common.util, no review.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This patch and the several to follow represent an attempt to break
up a very large patch from martin which moves much of the compiler's
typing infrastructure into the library so that we can utilize the same
machinery at runtime that we do at compile time.
Said attempt was not very successful, either at maintaining working
order while committing incrementally or at subdividing the patch into
distinct cohesive units. So let the record show that I tried.
-- some notes on the implementation --
This should not be judged as a finished work, but it's a necessary
opening step for implementing a reflection API without duplicating much
of the compiler.
The files in scala.reflect.common are destined for the library (not the
compiler, where they are now) but have not yet made the leap to manage
compatibility issues.
scala.reflect.generic is likely to be replaced by an alias to
scala.reflect.common.
The fate of scala.reflect.Code and its derived classes is yet to be
determined.
This first patch contains about the only delta against martin's patch,
which is to promote Chars into scala.reflect. Review by odersky.
|
| |
|
|
|
|
|
|
| |
Removing two more @bridge methods because they do not make sense
anymore. Review by extempore.
|
| |
|
|
|
|
|
|
|
|
|
| |
After numerous red herrings, I think I find my memory issues arise from
an $outer pointer: the inner class's innocuous seeming reference of a
hashmap has impressive consequences, especially in partest where it was
multiplied by the number of workers. Perhaps I will finally now see the
benefits I expect. No review.
|
|
|
|
|
|
|
| |
An ill-advised change to an equals method resulted in lots and lots and
lots of extra DirEntry/FileEntry instances. This should clear up recent
memory issues and put us nicely into the black overall. No review.
|
|
|
|
|
|
|
|
|
|
| |
but my theory is that ++ takes a by name argument, but doing a foldLeft
and using ++ to join creates a closure which loses the by-nameness. If
this theory is correct that's an ugly trap.
Not sure how I write a test against this sort of thing? Will take
pointers. For now, closes #4582, no review.
|
|
|
|
|
|
|
|
| |
Confusingly, I am restoring the method "getArchive" which has been gone
from trunk since February 2010 but is suddenly necessary for the sbt
build (as its "compiler-interface" won't compile without it.) I am sure
it all makes perfect sense. No review.
|
|
|
|
|
|
|
| |
Change Types#narrow to create an existential rather than a refinement
type, as the comment indicated was a desirable outcome. (Definitely)
review by odersky.
|
| |
|
|
|
|
|
| |
it's clearly time, but review by odersky.
|
|
|
|
|
|
| |
Added flatten to Option so Some(Some(x)).flatten returns Option[Int] and
not Iterable[Int]. No review.
|
| |
|
|
|
|
|
| |
Reverting the rest of it until I can look with clear eyes, no review.
|
| |
|