azalea/kotlin-fork
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
13,084 Commits 1 Branch 0 Tags
ddda69c0a6ee98a5a39f5c4c0c1dd7e4e9da2e28
Commit Graph

13084 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Erokhin Stanislav ddda69c0a6 JS Backend: hack nested object access 2013-12-02 19:56:12 +04:00
Alexander Udalov 97ad75a9f9 JS Backend: fix tests after object refactoring 2013-12-02 19:56:12 +04:00
Alexander Udalov ba693d3e38 Delete nested_object_name from binary format 
It's useless now: nested objects are stored along with classes
 2013-12-02 19:56:12 +04:00
Alexander Udalov 4526d96186 Delete JetScope.getObjectDescriptor() 2013-12-02 19:56:11 +04:00
Alexander Udalov 550df37056 Delete JetScope.getObjectDescriptors() 2013-12-02 19:56:11 +04:00
Alexander Udalov 31c84951dc Fix objects appearing in type position in completion 
After the object refactoring they're present as classes in scopes, not as
object descriptors
 2013-12-02 19:56:11 +04:00
Alexander Udalov 9e6cc829fb Delete WritableScopeImpl.addObjectDescriptor() 2013-12-02 19:56:10 +04:00
Alexander Udalov c2c4123f64 Don't store object descriptors separately in lazy member scope 
Object descriptors are refactored to be the usual classes, so the scope can't
contain any 'objects'
 2013-12-02 19:56:10 +04:00
Alexander Udalov 9890f2e5ac Simplify CodegenBinding.canHaveOuter(), make it work for objects 
This fixes CompileKotlinAgainstKotlinTest
 2013-12-02 19:56:10 +04:00
Alexander Udalov 90490b5c0c Delete OBJECT_DECLARATION_NAME slice and relevant usages 
JetObjectDeclarationName is almost useless now, since properties aren't created
for objects anymore
 2013-12-02 19:56:10 +04:00
Alexander Udalov 0792af5f85 JetObjectDeclarationName is no longer JetDeclaration 
Fix lazy resolve descriptor renderer tests after object refactoring
 2013-12-02 19:56:09 +04:00
Alexander Udalov f0e230cf23 Fix test data of recursive processor and renderer 
After object refactoring
 2013-12-02 19:56:09 +04:00
Alexander Udalov c6cfeec6fc Delete VariableDescriptorForObject 2013-12-02 19:56:09 +04:00
Alexander Udalov e3fffa6b0b Fix decompiled data for objects and enum entries 
Also prettify it for enum entries: output only names
 2013-12-02 19:56:09 +04:00
Alexander Udalov ce481b08d8 Fix CompileKotlinAgainstCustomBinariesTest 
The two cases are almost irrelevant now that objects aren't resolved to
properties anymore
 2013-12-02 19:56:08 +04:00
Alexander Udalov cb6c98d4d6 Delete OBJECT_PROPERTY kind from binary format 
We never create properties for objects anymore, except enum entries of Java
enums. Delete PropertyDescriptorForObjectImpl and its usages
 2013-12-02 19:56:08 +04:00
Alexander Udalov 42839fa061 Delete LoadKotlinCustomTest 
It's useless now because enums are loaded the same way from binaries as well as
from sources
 2013-12-02 19:56:08 +04:00
Alexander Udalov dd290bbeb9 Create classes when deserializing enum entries 2013-12-02 19:56:07 +04:00
Alexander Udalov bd5eee5cc2 Put enum entries into enum in lazy resolve 2013-12-02 19:56:07 +04:00
Alexander Udalov 5a4cc1e2fe Put enum entries into enum class in JDR 2013-12-02 19:56:07 +04:00
Alexander Udalov 60e7722a82 Enum entries are put into enum class, not its class object 2013-12-02 19:56:06 +04:00
Alexander Udalov a5d6d6719c EnumValue now has a ClassDescriptor for its enum entry 2013-12-02 19:56:06 +04:00
Alexander Udalov 83ef095093 Change enum entry resolution strategy in JDR 
Enum entry is now resolved into a class with a class object, which inherits
from the former class, as in the other parts of the compiler. Create a special
class EnumEntrySyntheticClassDescriptor which will be reused in deserialization
later
 2013-12-02 19:56:06 +04:00
Alexander Udalov 57678a5506 Fix AnnotationDescriptorResolveTest, objects are now classes in scopes 2013-12-02 19:56:06 +04:00
Alexander Udalov 6f930ef725 Support updated enum entry hierarchy in JVM codegen 2013-12-02 19:56:05 +04:00
Alexander Udalov edc37bd499 Change local objects resolution, delete LocalVariableDescriptorForObject 
Local objects are now resolved as global ones (with derived class objects)
 2013-12-02 19:56:05 +04:00
Alexander Udalov 2967b69c4a Update enum entry resolution strategy in lazy resolve 2013-12-02 19:56:05 +04:00
Alexander Udalov bf1ea5fb01 Drop NamespaceLikeBuilder.addObjectDescriptor 2013-12-02 19:56:05 +04:00
Alexander Udalov dd6cfcdc78 Resolve enum entries similar to objects in TypeHierarchyResolver 2013-12-02 19:56:04 +04:00
Alexander Udalov c04f63e157 Change object deserialization strategy 
Objects now have synthetic class objects in deserialization as well. "Class"
proto message now can have a synthetic class object proto within, if it's not
contained in any external class file, which is the case with objects. Drop
"class_object_present" field from binary format, since its value is equivalent
to the presence of "class_object" field
 2013-12-02 19:55:21 +04:00
Alexander Udalov 26bd1ff189 Change object resolution strategy in lazy resolve 2013-12-02 19:55:20 +04:00
Alexander Udalov ae116c6d9a Support updated object hierarchy in JVM codegen 2013-12-02 19:55:20 +04:00
Alexander Udalov 197eaa0b2e Drop BodiesResolveContext.getObjects() 
Objects are stored in the same place where classes are
 2013-12-02 19:55:20 +04:00
Alexander Udalov 6c74b10f6c Change object resolution strategy in TypeHierarchyResolver 
Instead of creating a synthetic property descriptor for objects, create a
synthetic class object in it and make it inherit from the object. This way all
of object's members will be accessible via its class object
 2013-12-02 19:55:20 +04:00
Alexander Udalov 18c59d5495 Treat anonymous objects as classes (ClassKind.CLASS) 
They're not objects per se, i.e. they're not singletons
 2013-12-02 19:55:19 +04:00
Alexander Udalov 27f832a713 BodiesResolveContext.getClasses()'s key is JetClassOrObject 
This has no effect right now, but is a preparation for the objects refactoring,
when BodiesResolveContext.getObjects() will be dropped
 2013-12-02 19:55:19 +04:00
Alexander Udalov 008e5f0b13 Don't serialize default primary constructors 
Drop "Callable primary_constructor" field, because it requires serializer to
serialize the fact that a class has a primary constructor. Default primary
constructors for objects etc. may not be serialized, because they can be
re-created trivially when they're loaded. There are three options: a class
doesn't have a primary constructor, a class has a default primary constructor,
a class has a custom primary constructor. This explains the change in
descriptors.proto
 2013-12-02 19:52:44 +04:00
Mikhael Bogdanov 606b3aa4a9 Test for local fun 2013-12-02 18:30:43 +04:00
Mikhael Bogdanov 59dbf75d80 Added test for sam 2013-12-02 18:30:43 +04:00
Mikhael Bogdanov 712dcaac91 Added tests for && and || 2013-12-02 18:30:43 +04:00
Mikhael Bogdanov 0728428814 Change renderer for descriptors 2013-12-02 18:30:43 +04:00
Mikhael Bogdanov 0b2c3efc40 Inline recursion diagnostic 2013-12-02 18:30:43 +04:00
Mikhael Bogdanov 9a9527906b Don't lost diagnostics on real elements when processing fake resolveCall 2013-12-02 18:30:42 +04:00
Mikhael Bogdanov 621a4002e4 Binary and unary expression support 2013-12-02 18:30:42 +04:00
Mikhael Bogdanov 431e61ef74 Beatify code 2013-12-02 18:30:42 +04:00
Mikhael Bogdanov 771501912e Comment 'inline' annotation that brokes build 2013-12-02 18:30:42 +04:00
Mikhael Bogdanov 1391676281 Support parenthesized, typed and labeled expressions 2013-12-02 18:30:42 +04:00
Mikhael Bogdanov ceda72e0ca Deserialization of function descriptors via DesirializedSimpleFunctionDescriptor 2013-12-02 18:30:42 +04:00
Mikhael Bogdanov 8533fd64ff Compiler inline on/off flag 2013-12-02 18:30:42 +04:00
Mikhael Bogdanov 4079735bf2 Diagnostic for nullable parameter of Function type 2013-12-02 18:30:42 +04:00