Commit message (Collapse) | Author | Age | Files | Lines | ||
---|---|---|---|---|---|---|
... | ||||||
| * | | New phase: collectEntryPoints | Dmitry Petrashko | 2014-12-16 | 2 | -69/+121 | |
| | | | | | | | | | | | | collects entry points and registers them in backend | |||||
| * | | Remove duplicate isAnonymousFunction | Dmitry Petrashko | 2014-12-16 | 2 | -4/+1 | |
| | | | ||||||
| * | | Move methods shared between Mixin and Backend to SymUtils. | Dmitry Petrashko | 2014-12-16 | 2 | -4/+20 | |
| | | | ||||||
| * | | Hack ClassFileParser to not treat Null$ or Nothing$ as modules. | Dmitry Petrashko | 2014-12-16 | 2 | -1/+6 | |
| | | | | | | | | | | | | They are classes. | |||||
| * | | Update to scala 2.11.4 | Dmitry Petrashko | 2014-12-16 | 1 | -1/+1 | |
| | | | ||||||
| * | | Erasure is on by default. | Dmitry Petrashko | 2014-12-16 | 1 | -30/+30 | |
| | | | | | | | | | | | | Remove historical enabling of erasure for some tests: its enabled by default now | |||||
| * | | Sharing backend between scalac and dotty. | Dmitry Petrashko | 2014-12-16 | 7 | -2/+1334 | |
| | | | | | | | | | | | | | | | Due to a lot of error and workarounds required for backend not regress in terms of speed, I guess the original idea was to high to be reached.. Nevertheless it was good to try. | |||||
* | | | Merge pull request #302 from dotty-staging/fix/returns | Dmitry Petrashko | 2014-12-17 | 5 | -31/+49 | |
|\ \ \ | | | | | | | | | Fix/returns | |||||
| * | | | Move isSuppressed check from Reporter to ConsoleReporter. | Martin Odersky | 2014-12-17 | 2 | -11/+12 | |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | isSuppressed forces computation of the error message; should be called only when emitting the error. TODO: Generalize so that other emitting reporters inherit the behavior. | |||||
| * | | | Label defs are not source methods. | Martin Odersky | 2014-12-17 | 1 | -1/+3 | |
| | | | | | | | | | | | | | | | | | | | | | | | | If this commit is made without the previous fixes to returns, we get errors in dotc_parsing (without the fix to polymorphic returns) and in dotc_core (without the fix in tree transform). | |||||
| * | | | Don't transform Return#from | Martin Odersky | 2014-12-17 | 1 | -1/+5 | |
| | | | | | | | | | | | | | | | | Explanation given in the comment. | |||||
| * | | | Fix returns from polymorphic methods | Martin Odersky | 2014-12-17 | 1 | -18/+29 | |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Return proto type was wrong; it was the type of the method (with PolyParams if that type is polymorphic), where it should have been the return type instantiated with the local type parameters. We now instantiate the prototype as required. Note the typing of return nodes is a bit more subtle than in scalac, because dotc trees are immutable, so the tree pointed at by the context is of no help - it does not have a return type which we could make use of. | |||||
* | | | | Merge pull request #301 from dotty-staging/fix/#300-literals | odersky | 2014-12-17 | 2 | -0/+32 | |
|\ \ \ \ | |/ / / |/| | | | Fix #300: Make literals have the same constants as their types. | |||||
| * | | | Fix Literalize: (Long(2):Any) == (Int(2): Any) but constants aren't. | Dmitry Petrashko | 2014-12-17 | 1 | -1/+1 | |
| | | | | ||||||
| * | | | Check not only value equality of constants in Literalize postcondition but ↵ | Dmitry Petrashko | 2014-12-17 | 1 | -3/+4 | |
| | | | | | | | | | | | | | | | | also type eqality | |||||
| * | | | Add checkPostCondition to Literalize | Dmitry Petrashko | 2014-12-17 | 1 | -0/+17 | |
| | | | | ||||||
| * | | | Fix #300: Make literals have the same constants as their types. | Martin Odersky | 2014-12-17 | 2 | -0/+14 | |
| | | | | ||||||
* | | | | Merge pull request #285 from dotty-staging/fix/catchNonFatal | Dmitry Petrashko | 2014-12-17 | 11 | -17/+32 | |
|\ \ \ \ | |/ / / |/| | | | Fix/catch non fatal | |||||
| * | | | Remove catch-rethrow in Driver | Dmitry Petrashko | 2014-12-17 | 1 | -2/+0 | |
| | | | | ||||||
| * | | | Made typedUnApply a hook that has to be overridden in a ReTyper. | Martin Odersky | 2014-12-15 | 1 | -8/+4 | |
| | | | | | | | | | | | | | | | | | | | | Change exhibits a problem in the transformers: pattern matching try's do not seem to be handled. | |||||
| * | | | Replace Throwable catches with NonFatal | Martin Odersky | 2014-12-15 | 9 | -17/+27 | |
| | | | | | | | | | | | | | | | | | | | | | | | | Previously, stackoverflows led to infinite loops because the catch immediately threw another stack overflow. Anyway, one should never catch Throwable. | |||||
| * | | | Add a typedUnApply method. | Martin Odersky | 2014-12-15 | 2 | -0/+11 | |
| | |/ | |/| | | | | | | | | | | The method is needed to be able to survive ReTypers between frontend and pattern matching. | |||||
* | | | Merge pull request #270 from dotty-staging/change/type-aliases | odersky | 2014-12-17 | 11 | -147/+144 | |
|\ \ \ | | | | | | | | | Change/type aliases | |||||
| * | | | Get rid of TypeBounds.onAlias | Martin Odersky | 2014-12-10 | 3 | -5/+2 | |
| | | | | | | | | | | | | | | | | This was the last bit that tested for the fragile `lo eq hi` condition. | |||||
| * | | | Revert TypeBounds.real to TypeBounds.apply | Martin Odersky | 2014-12-10 | 2 | -10/+10 | |
| | | | | | | | | | | | | | | | | | | | | TypeBounds.real served its purpose: Flushing out all problems with creating TypeBounds. | |||||
| * | | | Make derivedTypeBounds always create real TypeBounds. | Martin Odersky | 2014-12-10 | 1 | -1/+1 | |
| | | | | ||||||
| * | | | Refactor & and | on TypeBounds | Martin Odersky | 2014-12-10 | 1 | -20/+23 | |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | Generate Type aliases only when original type(s) were aliases. Also, TypeBounds.real never generates an alias. | |||||
| * | | | Distinguish calls to create real type bounds from alias type bounds. | Martin Odersky | 2014-12-10 | 4 | -22/+23 | |
| | | | | | | | | | | | | | | | | | | | | So far this affects only the calls, both methods are still the same. But their implementation will be changed next. | |||||
| * | | | Moved derivedTypeAlias method to TypeAlias | Martin Odersky | 2014-12-10 | 5 | -86/+77 | |
| | | | | ||||||
| * | | | Converted some TypeBounds pattern matches to TypeAlias matches. | Martin Odersky | 2014-12-10 | 4 | -15/+13 | |
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Replace all case TypeBounds(lo, hi) if lo eq hi => patterns with case TypeAlias(alias) => | |||||
| * | | | TypeAlias unapply accepts only Type Aliases | Martin Odersky | 2014-12-10 | 2 | -4/+6 | |
| | | | | ||||||
| * | | | Create TypeAlias subclass of TypeBouunds | Martin Odersky | 2014-12-10 | 4 | -31/+29 | |
| | | | | | | | | | | | | | | | | Specialize hash consing of type aliases. | |||||
| * | | | Add equality method to TypeBounds. | Martin Odersky | 2014-12-10 | 1 | -0/+7 | |
| | | | | | | | | | | | | | | | | | | | | Curiously, was missing before, which meant variance was ignored when hash-consing TypeBounds. | |||||
* | | | | Merge pull request #296 from dotty-staging/fix/#288-by-name-params-in-mixins | Dmitry Petrashko | 2014-12-16 | 3 | -6/+24 | |
|\ \ \ \ | | | | | | | | | | | Fix problem with by name params in mixed in methods. | |||||
| * | | | | Fix problem with by name params in mixed in methods. | Martin Odersky | 2014-12-16 | 3 | -6/+24 | |
| | |/ / | |/| | | | | | | | | | | Erasure choked on them before. | |||||
* | | | | Merge pull request #299 from dotty-staging/try-handling | Dmitry Petrashko | 2014-12-16 | 2 | -2/+15 | |
|\ \ \ \ | | | | | | | | | | | Try handling inside patternmatcher | |||||
| * | | | | Provide a simple implementation of transformTry for backend. | Dmitry Petrashko | 2014-12-16 | 1 | -1/+14 | |
| | | | | | | | | | | | | | | | | | | | | Will need to revisit it later in order to generate faster code. | |||||
| * | | | | Fix printing of Try nodes with non-empty cases. | Dmitry Petrashko | 2014-12-16 | 1 | -1/+1 | |
|/ / / / | ||||||
* | | | | Merge pull request #294 from samuelgruetter/tests-rewrite-2 | odersky | 2014-12-16 | 509 | -64/+125 | |
|\ \ \ \ | | | | | | | | | | | Run rewrite tool on tests (2) | |||||
| * | | | | move failing tests from tests/untried/pos to tests/pending/pos | Samuel Gruetter | 2014-12-16 | 494 | -0/+0 | |
| | | | | | ||||||
| * | | | | run srewrite version 88d3cd4126d135617a8189f0a855757d7e2ab806 on ↵ | Samuel Gruetter | 2014-12-16 | 35 | -57/+118 | |
| | |_|/ | |/| | | | | | | | | | | tests/untried/pos | |||||
* | | | | Merge pull request #282 from dotty-staging/fix/i0239-package-objects | odersky | 2014-12-16 | 5 | -9/+46 | |
|\ \ \ \ | |/ / / |/| | | | Fix #239 - handling of package objects | |||||
| * | | | Fix to asSeenFrom for package object members | Martin Odersky | 2014-12-15 | 2 | -0/+26 | |
| | | | | | | | | | | | | | | | | | | | | Like TypeAssigner, asSeenFrom needs to insert a package object if the prefix is a package but the class enclosing the type is not. | |||||
| * | | | Fix #239 - handling of package objects | Martin Odersky | 2014-12-13 | 3 | -9/+20 | |
| | | | | | | | | | | | | | | | | | | | | | | | | References to `.package` are now also inserted if the accessed member comes from a class inherited by a package object. | |||||
* | | | | Merge pull request #292 from dotty-staging/fix/#290-type-bind | Dmitry Petrashko | 2014-12-16 | 4 | -5/+29 | |
|\ \ \ \ | | | | | | | | | | | Fix/#290 type bind | |||||
| * | | | | Fix typechecking rules for Binds of type trees. | Martin Odersky | 2014-12-15 | 3 | -5/+27 | |
| | | | | | ||||||
| * | | | | Make Bind nodes type trees depending on bound name. | Martin Odersky | 2014-12-15 | 1 | -0/+2 | |
| | |/ / | |/| | | | | | | | | | | Bind nodes were always TermTrees, which is wrong. | |||||
* | | | | Merge pull request #293 from smarter/fix/multiple-refinements-warning | Dmitry Petrashko | 2014-12-16 | 2 | -3/+4 | |
|\ \ \ \ | |/ / / |/| | | | Avoid spurious warnings about forward references in refinements | |||||
| * | | | Avoid spurious warnings about forward references in refinements | Guillaume Martres | 2014-12-16 | 2 | -3/+4 | |
|/ / / | | | | | | | | | | | | | | | | | | | | | | The warning was triggered by cases like: class A type F = A { type T = Int; def f: T } Which is treated differently from the following which did not produce a warning: type F = A { type T = Int } { def f: T } | |||||
* | | | Merge pull request #291 from dotty-staging/fix-transform-bind | odersky | 2014-12-15 | 1 | -2/+2 | |
|\ \ \ | | | | | | | | | Fix Bind node contents not being transformed if phase didn't register transformBind. |