| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Implicit search detects likely cycles by looking at the stack of
open implicits and checking the same implicit appears twice, and
if the second occurrence is trying satisfy an implicit search for
a "dominant" type.
Originally, this condition immediately failed the entire implicit
search. However, since Scala 2.10, this mechanism has been refined to
continue searching after the first divergent implicit is detected.
If a second divergence is found, we fail immediately. If the followup
search fails, we report the first divergence. Otherwise, we
take the successful result.
This mechanism was originally built around exceptions. This proved
to be fragile, and was refactored in SI-7291 / accaa314 to instead
use the `Context.errors` to control the process.
But, since that change, the pattern of implicits in scalanlp/breeze
and Shapeless have been prone to reporting the divergent implicit
errors where they used to recover.
So long as we left the `DivergentImplictTypeError` that originates
from a nested implicit search in `context.errors`, we are unable to
successfully typecheck other candidates. This commit instead
stashes the first such error away in `DivergentImplicitRecovery`,
to clear the way for the alternative path to succeed.
We must retain any other divergent implicit errors, as witnessed by
test/files/neg/t2031.scala, which loops unless we retain divergent
implicit errors that we don't stash in `DivergentImplicitRecovery`.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
In SI-7944 / 251c2b93, we discovered that typechecking of an implicit
candidate could leave orphaned undetermined type parameters in the
implicit search context. This resulted in naked type parameters
leaking into implicit expansions. The fix seemed easy: just copy
any symbols from `implicitSearchContext.undetparams` to the enclosing
context (other than ones in `SearchResult#subst`).
However, the test case in this ticket reveals a subtle flaw in that
fix: `implicitSerachContext.undetparams` only contains the type params
from the most recently typechecked candidate!
Why? Implicit search uses the same context to typecheck all plausibly
compatible candidates. The typechecking itself is driven by
`typedImplicit1`.
Side note, that explains the heisenbug behaviour noted in the ticket:
Not *all* plausibly implicit candidates are typechecked. If
the current 'best' eligible candidate is more specific than the next
candidate, we can skip that altogether.Implicit search actually
exploits this for performance by ordering the candidates according to
usage statistics. This reordering, means that commenting out lines
elsewhere in the file changed the behaviour!
This commit simply stores the undet. tparams in the `SearchResult`,
where it is safe from the vaguries of typechecking other candidates.
That makes `Test1` and `Test2` with in the enclosed test case fail
uniformly, both with each other, and with an explicit call to
the view.
This is ostensibly a regression from 2.10.3. To get there,
we need an implicit search that has to infer `Nothing` in a covariant
position. In 2.10.3, we would just let the `G` out into the wild,
which did the right thing for the wrong reasons.
|
|\
| |
| | |
SI-4492 More informative error when class not found on classpath
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Position the error based on Select tree that failed to type check,
presumably due to an underlying MissingRequirementError, which has no position.
There are lots of other ways we could rewrap a MRE and supplement position info,
but that remains TODO. Jason's review comment is recorded in the code.
Also try to detect the case of a missing module and provide some advice,
as well as linking to the forthcoming 2.11 guide at
http://docs.scala-lang.org/overviews/core/scala-2.11.html.
|
|/
|
|
|
|
|
| |
Issue deprecation warning under -Xsource:2.10 so time
travelers can have an authentic deprecation experience before
finding that their unsound code no longer compiles in 2.11.
The relevant ticket to the soundness issue is SI-6566.
|
|
|
|
|
| |
`T*` rather than `<repeated>[T]`, as we alreday do for Scala repeated
parameters.
|
|\
| |
| | |
SI-8375 saner binary incompat errors for macros
|
| | |
|
| |
| |
| |
| |
| |
| | |
Inspired by Brian McKenna's RC1 migration experience, this commit improves
macro impl binding validation in order to provide more helpful diagnostic
for this quite frequent class of errors.
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
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.
|
|/
|
|
|
| |
This commit merely documents current (suboptimal) error message printed
for the code reported in SI-8372. The next commit will fix the problem.
|
|
|
|
|
|
|
| |
Previously mutable.ListMap was used with assumption that it preserved
order of inserted elements (but it doesn't). Surprisingly logic that
assumed ordered elements worked mosly fine on unordered ones. I guess
two bugs can cancel each other out.
|
| |
|
|\
| |
| | |
SI-6455 no longer rewrite .withFilter to .filter
|
| |
| |
| |
| |
| | |
This has been deprecated for two major releases now,
but `filter`'s still preferred over `withFilter` in the wild.
|
|\ \
| | |
| | | |
SI-1503 don't assume unsound type for ident/literal patterns
|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
The fix only kicks in under -Xfuture. We also warn under -Xlint.
What type should a variable bound to the value matched by a pattern have?
To avoid CCEs, it should be a type that's implied by the matching
semantics of the pattern.
Usually, the type implied by a pattern matching a certain value
is the pattern's type, because pattern matching implies instance-of checks.
However, Stable Identifier and Literal patterns are matched using `==`,
which does not imply a type for the binder that binds the matched value.
The change in type checking due to this fix is that programs that used to crash with a CCE
(because we blindly cast to the type of the pattern, which a `==` check does not imply)
now get a weaker type instead (and no cast). They may still type check, or they may not.
To compensate for this fix, change `case x@Foo => x` to `case x: Foo.type => x`,
if it's important that `x` have type `Foo.type`.
See also:
- SI-4577: matching of singleton type patterns uses `eq`,
not `==` (so that the types are not a lie).
- SI-5024: patmat strips unused bindings, but affects semantics
|
|\ \ \
| | | |
| | | | |
SI-8321 bundles can't be whitebox
|
| | | |
| | | |
| | | |
| | | |
| | | | |
Given the recent glaring oversight in macro bundles, I have to have more
tests in order to make sure that things are going to work as they should.
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
At the moment, bundle selection mechanism is pretty picky. If a candidate
bundle's parameter isn't either blackbox.Context, whitebox.Context or
PrefixType refinement thereof, then it's not a bundle and the user
will get a generic error.
However we can be a bit more helpful and admit classes that are almost
like bundles (looksLikeMacroBundleType), have them fail isMacroBundleType,
and then emit a much prettier error message to the user that would tell them
that bundles must be monomorphic and their sole parameter should not just
be any subtype of blackbox.Context or whitebox.Context.
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
It's not like they were inducing bugs, but I can't see how polymorphism
can be useful for macro bundles, hence imho it's better to reduce the
number of degrees of freedom of the system.
|
| | | |
| | | |
| | | |
| | | |
| | | | |
Vanilla macros only allow blackbox.Context, whitebox.Context and
PrefixType refinements thereof. Bundles should behave in the same way.
|
| | | |
| | | |
| | | |
| | | |
| | | | |
whitebox.Context <: blackbox.Context, so in order to check for blackboxity
it's not enough to check whether the context used is <: blackbox.Context.
|
|\ \ \ \
| |/ / /
|/| | | |
SI-5479 deprecate DelayedInit outside of App
|
| | | | |
|
| | |/
| |/|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
DelayedInit's semantics are way too surprising.
For example, it delays initialization of fields,
so that fields on objects that extend `App`
(which `extends DelayedInit`) are not initialized
until the `main` method is called.
For more details and a proposed alternative,
see https://issues.scala-lang.org/browse/SI-4330?jql=labels%20%3D%20delayedinit%20AND%20resolution%20%3D%20unresolved.
Support for `App` will continue -- we'll special case it.
|
|\ \ \
| | | |
| | | | |
Small Predef cleanup
|
| | |/
| |/|
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
To support the established pattern for disabling it for
an compilation unit.
Update scaladoc's knowledge of our "typeclasses".
Leave a `private[scala]` version of `StringAdd` (public in bytecode)
to ensure binary compatibility with 2.11.0-M8 for partest.
|
|\ \ \
| | | |
| | | | |
SI-6675 deprecation warning for auto-tupling in patterns
|
| | |/
| |/|
| | |
| | |
| | | |
NOTE: when the deprecation warning becomes an error,
SI-6111 must become a `won't fix`
|
|/ /
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
1. Rename cardinality into rank. Shorter word, easier to
understand, more appropriate in our context.
2. Previously we called any dollar substitution splicing
but this is not consistent with Scheme where splicing
is substitution with non-zero rank.
So now $foo is unquoting and ..$foo and ...$foo is
unquote splicing or just splicing. Correspondingly
splicee becomes unquotee.
3. Rename si7980 test into t7980
|
|\ \
| |/
|/| |
SI-8063 and its seventy friends
|
| |
| |
| |
| |
| |
| | |
Highlights the dilemma with rich type members in the cake that no longer
exists. One used to have to choose between overloading or patmat/extmeth
friendliness, but couldn't have both. Thanks to retronym we can have it all.
|
| |\ |
|
| |\ \
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Conflicts:
src/compiler/scala/reflect/macros/compiler/Resolvers.scala
src/compiler/scala/reflect/macros/contexts/Typers.scala
src/compiler/scala/tools/reflect/ToolBoxFactory.scala
src/reflect/scala/reflect/api/BuildUtils.scala
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
It’s almost 1am, so I’m only scratching the surface, mechanistically
applying the renames that I’ve written down in my notebook:
* typeSignature => info
* declarations => decls
* nme/tpnme => termNames/typeNames
* paramss => paramLists
* allOverriddenSymbols => overrides
Some explanation is in order so that I don’t get crucified :)
1) No information loss happens when abbreviating `typeSignature` and `declarations`.
We already have contractions in a number of our public APIs (e.g. `typeParams`),
and I think it’s fine to shorten words as long as people can understand
the shortened versions without a background in scalac.
2) I agree with Simon that `nme` and `tpnme` are cryptic. I think it would
be thoughtful of us to provide newcomers with better names. To offset
the increase in mouthfulness, I’ve moved `MethodSymbol.isConstructor`
to `Symbol.isConstructor`, which covers the most popular use case for nme’s.
3) I also agree that putting `paramss` is a lot to ask of our users.
The double-“s” convention is very neat, but let’s admit that it’s just
weird for the newcomers. I think `paramLists` is a good compromise here.
4) `allOverriddenSymbols` is my personal complaint. I think it’s a mouthful
and a shorter name would be a much better fit for the public API.
|
| |\ \ \ |
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
I have finally overcome my fear of positions and got to cleaning up its
public interface.
Apparently it isn’t so bad, since there’s a sane core of methods (thanks
to whoever wrote the comments to internal#Position):
1) Checks to distinguish offsets, opaque ranges and transparent ranges
2) Essentials that inclide start, point, end and source
3) Factories that create new positions based on existing ones
It looks like methods from the 3rd group are exactly what we’ve been looking
for in SI-6931, so we have nothing to add in this commit.
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
Continuing in the direction set by the parent commit, this commit
rephrases some more usages of `local` in names and comments in typer.
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
Pretty much everything seems like it's intended for internal use, so I moved it to a package-private object.
Split toParArray out and put it in an implicit class. Added ability to .toParArray from Array and String also.
Added test to verify implicits are gone.
|
| |_|_|/
|/| | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Experience building open source projects like Specs that use
`-Xlint` suggests that this warning is too noisy to lump in
with the others.
We are lacking in more fine-grained control of these things, so
simply turning of `-Xlint` in favour of its underlying `-Y` options
ends up *losing* some other important warnings that are predicated
directly on `-Xlint`.
Furthermore, bug reports against M8, SI-7707 SI-7712, show that
unused private/local warnings, while far less noisy, are still
in need of polish.
This commit moves these warnings to a pair of new -Y options,
neither of which is part of `-Xlint`..
Let's ask people to opt in for 2.11, and as it stabilizes, we can
consider adding it to Xlint (or the desirable evolution of that)
in the next release.
|
| |_|/
|/| |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
This reverts commit a02e053a5dec134f7c7dc53a2c1091039218237d.
That commit lead to an error compiling Specs2:
[info] [warn] /localhome/jenkinsdbuild/workspace/Community-2.11.x-retronym/dbuild-0.7.1-M1/target-0.7.1-M1/project-builds/specs2-aaa8091b47a34817ca90134ace8b09a9e0f854e9/core/src/test/scala/org/specs2/text/EditDistanceSpec.scala:6: Unused import
[info] [warn] import DiffShortener._
[info] [warn] ^
[info] [error] /localhome/jenkinsdbuild/workspace/Community-2.11.x-retronym/dbuild-0.7.1-M1/target-0.7.1-M1/project-builds/specs2-aaa8091b47a34817ca90134ace8b09a9e0f854e9/core/src/test/scala/org/specs2/text/LinesContentDifferenceSpec.scala:7: exception during macro expansion:
[info] [error] java.lang.UnsupportedOperationException: Position.point on NoPosition
[info] [error] at scala.reflect.internal.util.Position.fail(Position.scala:53)
[info] [error] at scala.reflect.internal.util.UndefinedPosition.point(Position.scala:131)
[info] [error] at scala.reflect.internal.util.UndefinedPosition.point(Position.scala:126)
[info] [error] at org.specs2.reflect.Macros$.sourceOf(Macros.scala:25)
[info] [error] at org.specs2.reflect.Macros$.stringExpr(Macros.scala:19)
|
|\ \ \
| | | |
| | | | |
SI-5920 enables default and named args in macros
|
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
When producing an initial spec for macros two years ago, we sort of glossed
over named/default arguments in macro applications, leaving them for future work.
Once the aforementioned future has come, I’ve made several attempts at
making things operational (e.g. last summer), but it’s always been unclear
how to marry the quite complex desugaring that tryNamesDefaults performs
with the expectations of macro programmers to see unsugared trees
in macro impl parameters.
Here’s the list of problems that arise when trying to encode named/default
arguments of macro applications:
1) When inside macro impls we don’t really care about synthetic vals
that are typically introduced to preserve evaluation order in non-positional
method applications. When we inline those synthetics, we lose information
about evaluation order, which is something that we wouldn’t like to lose
in the general case.
2) More importantly, it’s also not very exciting to see invocations of
default getters that stand for unspecified default arguments. Ideally,
we would like to provide macro programmers with right-hand sides of those
default getters, but that is: a) impossible in the current implementation
of default parameters, b) would anyway bring scoping problems that we’re
not ready to deal with just yet.
Being constantly unhappy with potential solutions to the aforementioned
problems, I’ve been unable to nail this down until the last weekend,
when I realized that: 1) even though we can’t express potential twists in
evaluation order within linearly ordered macro impl params, we can use
c.macroApplication to store all the named arguments we want, 2) even though
we can’t get exactly what we want for default arguments, we can represent
them with EmptyTree’s, which is not ideal, but pretty workable. That’s
what has been put into life in this commit.
As a pleasant side-effect, now the macro engine doesn’t have to reinvent
the wheel wrt reporting errors about insufficient arg or arglist count.
Since this logic is intertwined with the tryNamesDefaults desugaring,
we previously couldn’t make use of it and had to roll our own logic
that checked that the number of arguments and parameters of macro applications
correspond to each other. Now it’s all deduplicated and consistent.
|
|\ \ \ \
| | | | |
| | | | | |
SI-8270 unconfuses bundles and vanilla macros
|
| | |/ /
| |/| |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
This fixes a mistake in macro impl ref typechecking that used to have
an heuristic to figure out whether it looks at a bundle method ref or at
a vanilla object method ref. Under some circumstances the heuristic could
fail, and then the macro engine would reject perfectly good macro impls.
Now every macro impl ref is typechecked twice - once as a bundle method ref
and once as a vanilla object method ref. Results are then analyzed,
checked against ambiguities (which are now correctly reported instead
of incorrectly prioritizing towards bundles) and delivered to the macro
engine.
The only heuristic left in place is the one that's used to report errors.
If both bundle and vanilla typechecks fail, then if a bundle candidate
looks sufficiently similar to a bundle, a bundle typecheck error is reported
providing some common bundle definition hints.
|
|\ \ \ \
| | | | |
| | | | | |
SI-3452 Correct Java generic signatures for mixins, static forwarders
|
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
[Parts of this patch and some of the commentary are from @paulp]
This took me so long to figure out I can't even tell you. Partly because
there were two different bugs, one which only arose for trait forwarders
and one for mirror class forwarders, and every time I'd make one set
of tests work another set would start failing. The runtime failures
associated with these bugs were fairly well hidden because you usually
have to go through java to encounter them: scala doesn't pay that much
attention to generic signatures, so they can be wrong and scala might still
generate correct code. But java is not so lucky.
Bug #1)
During mixin composition, classes which extend traits receive forwarders
to the implementations. An attempt was made to give these the correct
info (in method "cloneBeforeErasure") but it was prone to giving
the wrong answer, because: the key attribute which the forwarder
must capture is what the underlying method will erase to *where the
implementation is*, not how it appears to the class which contains it.
That means the signature of the forwarder must be no more precise than
the signature of the inherited implementation unless additional measures
will be taken.
This subtle difference will put on an unsubtle show for you in test
run/t3452.scala.
trait C[T]
trait Search[M] { def search(input: M): C[Int] = null }
object StringSearch extends Search[String] { }
StringSearch.search("test"); // java
// java.lang.NoSuchMethodError: StringSearch.search(Ljava/lang/String;)LC;
The principled thing to do here would be to create a pair of
methods in the host class: a mixin forwarder with the erased
signature `(String)C[Int]`, and a bridge method with the same
erased signature as the trait interface facet.
But, this turns out to be pretty hard to retrofit onto the
current setup of Mixin and Erasure, mostly due to the fact
that mixin happens after erasure which has already taken
care of bridging.
For a future, release, we should try to move all bridging
after mixin, and pursue this approach. But for now, what can
we do about `LinkageError`s for Java clients?
This commit simply checks if the pre-erasure method signature
that we generate for the trait forward erases identically to
that of the interface method. If so, we can be precise. If not,
we emit the erased signature as the generic signature.
Bug #2) The same principle is at work, at a different location.
During genjvm, objects without declared companion classes
are given static forwarders in the corresponding class, e.g.
object Foo { def bar = 5 }
which creates these classes (taking minor liberties):
class Foo$ { static val MODULE$ = new Foo$ ; def bar = 5 }
class Foo { static def bar = Foo$.MODULE$.bar }
In generating these, genjvm circumvented the usual process whereby one
creates a symbol and gives it an info, preferring to target the bytecode
directly. However generic signatures are calculated from symbol info
(in this case reusing the info from the module class.) Lacking even the
attempt which was being made in mixin to "clone before erasure", we
would have runtime failures of this kind:
abstract class Foo {
type T
def f(x: T): List[T] = List()
}
object Bar extends Foo { type T = String }
Bar.f(""); // java
// java.lang.NoSuchMethodError: Bar.f(Ljava/lang/String;)Lscala/collection/immutable/List;
Before/after this commit:
< signature f (Ljava/lang/String;)Lscala/collection/immutable/List<Ljava/lang/String;>;
---
> signature f (Ljava/lang/Object;)Lscala/collection/immutable/List<Ljava/lang/Object;>;
This takes the warning count for compiling collections under
`-Ycheck:jvm` from 1521 to 26.
|
|\ \ \ \ \
| | | | | |
| | | | | | |
SI-8266 Deprecate octal escapes in f-interpolator
|