| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
| |
This reverts commit 9276a1205f74fdec74206209712831913e93f359.
The change is not binary compatible, See discussion on SI-8899. Making
filterImpl non-private changes its call-sites (within TraversableLike)
from INVOKESTATIC to INVOKEINTERFACE. Subclasses of TraversableLike
compiled before this change don't have a mixin for filterImpl.
|
|
|
|
|
|
|
|
|
|
| |
Used the standard method of sending out two iterators, one twice as fast as the others, to avoid hanging on .force, .hasDefiniteSize, and .addString.
.addString appends a "..." as the last element if it detects a cycle. It knows how to print the cycle length, but there's no good way to specify what you want right now, so it's not used.
Added tests in t8680 that verify that cyclic streams give the expected results.
Added to whitelist names of methods formerly used for recursion (now looping).
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The obvious fix, overriding `filterNot` in Stream, is not binary
compatible, see https://github.com/scala/scala/pull/3925
Instead, this makes `filterImpl` in TaversableLike private[scala],
which allows overriding it in Stream. The corresponding mima-failures
can be whitelisted, as the changes are only to private[scala].
In 2.12.x we can remove the override of `filter` in Stream, but in
2.11.x this is not binary compatible.
Eventually we'd also like to make filter / filterNot in
TraversableLike final, but that's not source compatible, so it cannot
be done in 2.12.x.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
We can copy the hash table from the parent scope, rather
than constructing it from scratch.
This takes us to:
% rm -rf /tmp/pkg; (for i in {1..50}; do for j in {1..100}; do echo "package pkg { class A_${i}_${j}___(val a: Int, val b: Int) }"; done; done) > sandbox/A1.scala && time qbin/scalac -Ybackend:GenASM -J-Xmx1G -J-XX:MaxPermSize=400M -d /tmp sandbox/A1.scala;
real 0m19.639s // head~1 was 0m35.662s
user 0m41.683s // head~1 was 0m58.275s
sys 0m1.886s
In more detail, this commit:
- Removes the unused `fingerprint` constructor parameter from
scopes. This is a remnant from a previous optimization attempt
- Leave only one constructor on Scope which creates an empty
scope
- Update the factory method, `newNestedScope`, to copy the hash
table from the parent if it exists. We can rely on the invariant
that `outer.hashTable != null || outer.size < MIN_HASH)`, so we
don't need `if (size >= MIN_HASH) createHash()` anymore. This
used to be needed in `Scope#<init>`, which accepted an aribitrary
`initElems: ScopeEntry`.
- Update subclasses and factories of `Scope` in runtime reflection
to accomodate the change. Pleasingly, we could actually remove
the override for `newNestedScope`.
- Unit tests the functionality I'm touching
|
|
|
|
|
|
|
|
|
|
|
|
| |
Due to the fact that all TypTrees are transformed into TypeTrees
during typechecking one couldn't treat typed type trees in the same
way as they treat untyped type trees.
This change implements support for pattern matching of TypeTrees as their
corresponding TypTree equivalent using tree preserved in the original.
The implementation itself is a trivial wrapping of regular TypTree
extractors into MaybeTypeTreeOriginal.
|
|
|
|
|
|
|
|
|
| |
I just realized that our tag-based shortcuts were incomplete, because
they only work with root mirrors (doing just u.typeTag[T].tpe means that
the type is going to be resolved in u.rootMirror because that's the default).
This commit fixes this oversight. I'm hoping for 2.11.0-RC3, but also feel
free to reschedule to 2.12.0-M1 if it becomes clear that RC3 isn't happening.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This liftable hasn't been originally included in the set of standard
liftables due to following contradiction:
1. On one hand we can have identity lifting that seems to be quite
consistent with regular unquoting:
q"..${List(1,2)}" <==> q"1; 2"
q"${List(1,2)}" <==> q"s.c.i.List(1, 2)"
q"..${List(q"a", q"b")}” <==> q"a; b"
q"${List(q"a", q"b")}" <==> q"s.c.i.List(a, b)"
This is also consistent with how lisp unquoting works although they
get lifting for free thanks to homoiconicity:
// scala
scala> val x = List(q"a", q"b); q"f($x)"
q"f(s.c.i.List(a, b))"
// scheme
> (let [(x (list a b))] `(f ,x))
'(f (list a b))
2. On the other hand lifting is an operation that converts a value into
a code that when evaluated turns into the same value. In this sense
Liftable[Tree] means reification of a tree into a tree that
represents it, i.e.:
q"${List(q"a", q"b")}"
<==>
q"""s.c.i.List(Ident(TermName("a")), Ident(TermName("b")))"""
But I belive that such lifting will be very confusing for everyone
other than a few very advanced users.
This commit introduces the first option as a default Liftable for trees.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Previously one could match a partial function with match quasiquote:
scala> val q"$scrutinee match { case ..$cases }" = q"{ case Foo => Bar
}"
scrutinee: universe.Tree = <empty>
cases: List[universe.CaseDef] = List(case Foo => Bar)
This was quite annoying as it leaked encoding of partial functions as
Match trees with empty tree in place of scrutinee.
This commit make sure that matches and partial functions are disjoint
and don't match one another (while preserving original encoding under
the hood out of sight of the end user.)
|
|
|
|
|
|
| |
Due to tree re-use it used to be the fact that type quasiquotes could
match term trees. This commit makes sure selections and applied type and
type applied are all non-overlapping between q and tq.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The b8a76f688c6ce2a4c305da064303bb46b53be875 introduced ArrayOps.{unzip,
unzip3} methods. Both of those methods have ClassTags as context bounds on
their type parameters so they can create (and return) instances of Arrays.
The type inference for those methods is supposed to be guided by
implicit evidence that T <: (T1, T2) (or T <: (T1, T2, T3) in unzip3 case).
However, context bounds are desugared into implicit parameters that
prepended in front of implicit parameters declared in source code. That
means the implicit evidence won't have a chance to guide type inference
because it comes as last implicit parameter.
This commit desugars context bounds and puts them at the end of implicit
parameter list. This way type inference is guided properly and we get
expected compiler errors for missing class tags.
The change to parameters order breaks binary compatibility with respect
to 2.11.0-RC1. I added filters to our binary compatibility configuration
files. We can get rid of them as soon as 2.11.0 is out.
|
|
Changes in the package scala.reflect.internals are ignored.
|