| Commit message (Collapse) | Author | Age | Files | Lines |
... | |
| |
|
|
|
|
|
| |
Changed class name from `Main` to `Test` in a failing test case.
|
|
|
|
|
| |
Review by dragos.
|
| |
|
|
|
|
|
|
|
| |
Renamed some methods in icode exception handlers for clarification,
fixed finalizers for 'return' inside a catch block. closes #4240. review
by rytz.
|
|
|
|
|
|
|
| |
Reverting some recent signature code and temporarily disabling a test so
I can work this out without shattering the tranquility of the build. No
review.
|
| |
|
|
|
|
|
|
| |
One still broken. (Very sorry, all laptop and no servers makes jack a
dull tester.) No review.
|
|
|
|
|
|
|
|
| |
Seem to have made a breakthrough with signature correctness. Most of
the remaining troubles were due to inherited members and their foreign
identifiers, but I think I found a simple way to make everyone happy.
Closes #4238, review by moors.
|
|
|
|
|
|
|
| |
Glad to see there was a test case; but we have to push harder when
writing tests to include the key characteristics of the function being
tested. No review.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Another lap around the track with generic signatures. At the root of the
issue reported in #4214 is our old friend (fondly remembered from the
days of primitive equality) boxed/primitive unification.
// scala
trait T[A] {
def f(): A
}
// the generic signature spec doesn't allow for parameterizing
// on primitive types, so this cannot remain Char. However
// translating it to Character, as was done, also has issues.
class C extends T[Char] {
def f(): Char = 'a'
}
// Note that neither of the signatures for f, the implementation // or
the bridge method, matches the type parameter. Generic interfaces in
class: T<java.lang.Character> Generic signatures: public char C.f()
public java.lang.Object C.f()
After this commit, primitive type parameters are translated into Object
instead of the boxed type. It was martin's idea, so no review. Closes
#4214.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Long overdue rehaul of presentation compiler tests. Added support for
options file, and switched simple-tests to use the STARR library on the
test classpath.
Put any options you need passed to the presentation compiler in a file
with the same name as the test, ending in '.opts'. Currently it's used
to set the -bootclasspath for simple-tests. Check the contents of
simple-tests.opts for all the options.
IMPORTANT: when committing a new STARR, you need to check this test.
Most likely, you *will* break it otherwise.
review by extempore
|
|
|
|
|
|
|
| |
Break 9 tests, fix 9 tests, realize you broke the test you checked in
between +9 and -9. Fix that and generalize the repl tests a little. I
think we're all the way there. No review.
|
| |
|
|
|
|
|
|
|
|
|
| |
Altered a check for impl classes to look at the name instead of the
flags, since the flags are wrong. Obviously we should fix the flags or
otherwise better address. Note that the test is well intentioned but
doesn't actually fail with older versions; the crash is not easy to
reproduce even when I simulate the whole repl. Review by dragos.
|
|
|
|
|
| |
No review.
|
|
|
|
|
| |
about how to obtain a String and how not to. Closes #4297, no review.
|
|
|
|
|
|
|
| |
lazy val calls shouldn't be included in pure expressions in general.
closes 4047. fixed also bug in optimizer (was eliminating a.foo call in
the test). review by dragos
|
|
|
|
|
| |
in future attempts. Closes #4291, references #4214, no review.
|
|
|
|
|
| |
(only to make you aware of the commit message in r24263.)
|
|
|
|
|
|
| |
Closes #4202 and removes empty TreeHashMap (superseded by HashMap
implementation). No review
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
default. The -toolcp option now works. -D and -J options are passed
as arguments to the runner program as well as to the underlying JVM,
because they may be needed for such things as fsc talking to a server
compiler. Added a new Settings type for -D and -J options. Deleted some
dead settings code.
There are dummy settings for -toolcp and -nobootcp so they can be seen
in the help output, which needs a big cleanup.
No review.
|
| |
|
|
|
|
|
|
| |
changed test output and comment, because mixed in fields and methods no
longer get a generic signature. Review by extempore.
|
| |
|
|
|
|
|
|
|
|
| |
Fixed a bug where newly introduced type parameters didn't have all the
substitutions done correctly. Fixes #4266.
No review.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Moved unlift to the Function companion object, which might have been
better in the first place. Had to make a minor change to genprod, and
then I couldn't escape that unscathed. Finished the not very complete
undertaking I found there to update the scaladoc. Lots of little changes
to the generated text and code.
I changed genprod to only put a unique stamp on Function0 so we can stop
having a 100 file diff everytime an i is dotted somewhere.
Closes #3825, no review.
|
|
|
|
|
|
|
|
| |
Observed that some historical issues with package objects no longer seem
so issuey. In the interests of keeping the arbitrary limitations to a
minimum, re-enabled case classes in package objects (see #2130, #3437.)
Closes #3437, review by odersky.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Based on the frequency with which I hear questions about it and similar,
this error message assumes too much understanding.
scala> scala.collection.mutable.MultiMap(1, 2, 3) <console>:8: error:
value MultiMap is not a member of package scala.collection.mutable
Now it says:
scala> scala.collection.mutable.MultiMap(1, 2, 3) <console>:8: error:
object MultiMap is not a member of package scala.collection.mutable
Note: trait MultiMap exists, but it has no companion object.
No review.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
Rediscovering that transpose sometimes throws an exception on irregular
lists and sometimes returns an irregular result (depending on whether
the first collection is the longest or not) indicated that this needs a
better resolution than the status quo. Determination: as long as we're
throwing an exception on any invalid input, we should throw an exception
on all invalid input, so that's what it does now. I postpone any attempt
to offer a variation accepting a "hole value". Closes #3597, review by
community.
|
| |
|
| |
|
|
|
|
|
| |
No review.
|
|
|
|
|
|
|
|
| |
Revert "Added missing string prefixes for names of map and set
collection classes." and related commits.
No review.
|
|
|
|
|
|
|
|
| |
Another fix for a broken test case involving new set and map string
reps.
No review.
|
|
|
|
|
| |
Fixing tests depending on map and set string representations. No review.
|
|
|
|
|
| |
No review.
|
| |
|
|
|
|
|
|
|
|
| |
Added new setting -Ygen-javap, which takes a directory as an argument
and generates javap output for every generated classfile. There is still
a lot of noise in a diff: still todo is postprocessing the output to
remove diff-noise-generating numbers and such. No review.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
I chased a lot of ghosts before finding the real culprit for why partest
failures have been unfailing.
Process(Seq("bash", "-c", "exit 42")) ! // 42 Process(Seq("bash",
"-c", "exit 42")) #> logFile ! // 0
That behavior is not yet fixed, but I altered partest not to use #> and
fixed the test which should have been failing but wasn't. Closes #4227,
no review.
|
|
|
|
|
|
|
|
| |
It seems to arise from the changes to inner objects which make them look
like lazy vals. I am not sure this is the right change but at least
it's only about a six character diff to review. Closes #4188, review by
odersky.
|
|
|
|
|
|
|
|
|
| |
Created invisible setting to expose empty package members outside of the
empty package, so the repl can use packages without going blind to the
empty package. This commit also eliminates a deadlock which hits when
the compiler starts logging before a lazy val has pulled its head from
the noose. Closes #4228, review by odersky.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
reinstate the disabled test.
COMMITTERS: the test is very fragile. If you don't run the whole test
suite (it's the very last one which runs!) then you broke the build.
Upon your toes you should be.
For anyone unaware, test/partest --update-check path/to/test will
magically pass any test and update the checkfile to resemble whatever
output it just produced. Incautious use of this feature may negatively
impact the mood of certain extempores.
No review.
|
|
|
|
|
|
|
|
|
|
| |
After seeing the 481 line diff I had to apply to bring the checkfile up
to date, I concluded this test cannot be in the general pool: this is
the third build break in the last few days, because it changes output
like some people change socks. (Hygenic people.) Sorry to move it to
disabled iulian, but I don't know what else to do with it on short
notice. Review by dragos.
|
|
|
|
|
| |
No review
|
|
|
|
|
|
| |
Applied a few more fixes for specialization and added a new instrumented
jar. Review by dragos.
|