| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
| |
Get rid of GenJVM and everything that refers to it.
Also get rid of GenAndroid since it's dead code that refers to GenJVM.
|
| |
|
|
|
|
|
|
| |
This commit provides the new sha1 codes of the new STARR.
Moreover, it replaces the implementation of StringContext.f
to `macro ???`. The implementation is magically hardwired into
`scala.tools.reflect.MacroImplementations.macro_StringInterpolation_f`
by the new STARR.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The problem of emitSWITCH not handling empty
keys popped up when I tried to implement
unfolding of pattern alternatives in genicode
instead of in typers/explicitouter.
This change makes perfect sense in isolation
as bytecode allows LOOKUPSWITCHes that have
only default case. I actually verified that
this kind of bytecode is generated by javac
when one has switch statement with only
default case defined.
Review by @paulp or @dragos.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
A remedy for an IllegalAccessError where generated bytecode referred to
an inaccessible type. Closes SI-1430.
Bonus materials:
- tore out all the invokedynamic support. The shipped jdk7
implementation shows limited resemblance to the one this was written
against; the code mostly serves to distract. (I think I could get
invokedynamic working pretty quickly, except that it would
mean having a codebase for java7 and one for 5-6, which is not a yak
I wish to shave today.)
- gave NullClass and NothingClass objects of their own, which
allowed a nice polymorphic simplification of isSubClass, plus a
couple other streamlinings.
|
| |
|
|
|
|
| |
Continuations suddenly needs fjbg.jar on its classpath due to transitive
dependency, no review.
|
| |
|
|
|
|
| |
Error reporting when the generated code size exceeds JVM limits (65,535
bytes per method). Closed #4573. review by extempore.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
IMPORTANT NOTE
Modifying code of the FJBG library is a very sensitive task since its
write operations are eg. used by the Scala compiler. This SVN commit
affects (should affect) only the read operations (which did never
work) of the FJBG library.
Changes include:
- fixed several killer bugs in the bytecode reader (see below).
- added missing input stream constructors in several Attribute
classes (eg. JBootstrapInvokeDynamic, JEnclosingMethodAttribute,
JInnerClassesAttribute) together with the corresponding
class registrations in class JAttributeFactory. - added
classes JExceptionsTable, JLocalVariableTableAttribute and
JStackMapTableAttribute (including corresponding factory methods in
class FJBGContext). - overriden method toString in most classes to
support output format similar to javap (see below). - did some code
cleanup (tabs, etc..).
Bug fixes include:
- fixed incorrect stream.readInt() in JSourceFileAttribute
constructor. - fixed missing code.lineNumbers initialization in
JLineNumberTableAttribute constructor. - fixed incorrect code in class
util.ByteArray constructor (stream). - added method setCode in class
JMethod to link them together.
Output of decoded bytecode:
The added toString() methods return javap-like formatted strings for
the decoded data, eg. for the LocalVariableTable attribute you get
the following output: ... LocalVariableTable: Start Length Slot Name
Signature 0 6 0 this LTest$; 0 6 1 args [Ljava/lang/String;
Executing the following Java code will produce decoded bytecode in a
format similar to the command line "javap -p -v -classpath <cpath>
<classes>":
static void printClass(String filePath) throws Exception {
FJBGContext fjbgContext = new FJBGContext(49, 0);
DataInputStream in = new DataInputStream(new FileInputStream(filePath));
JClass jclass = fjbgContext.JClass(in);
System.out.println(jclass);
in.close();
}
Mainly for providing a demonstrator of the FJBG reader we added to
the FJBG library the main class ch.epfl.lamp.fjbg.Main which behaves
similarly to javap, the class file disassembler of the J2SE SDK.
For instance the following commands produce (more or less) the same
output:
~$ scala ch.epfl.lamp.fjbg.Main -classpath classes 'Test javap -p -v$'
~$ -classpath classes 'Test $'
In several cases fjbg.Main will provide more information for access
flags (eg. brigde methods) and class file attributes (eg. enclosing
methods).
TODO
- integration of FJBGContext.JLocalVariableTableAttribute into the JVM
backend of the Scala compiler (cleaner handling of local variables).
- The source code of the FJBG library is currently generated using
the compiler option "-source 1.4"; moving to source release 1.5 would
allow further code improvements like: List --> List<T> (cast removals)
StringBuffer --> StringBuilder (faster implementation) /*@Override*/ -->
@Override
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
|
| |
|
|
|
|
|
|
| |
Fix InnerClasses attribute: anonymous classes don't have an outer
name. EnclosingMethod is correctly generated. Fixed isAnonymousClass
definition. Updated test that depends on anonymous inner class names.
Closes (again) #3249, references #2749. review by odersky,extempore.
|
| |
|
|
|
|
|
|
| |
This should fix java signatures when they refer to
method type parameters. I unrolled Adriaans previous fix
for #3249, as this one is more general. Closes #3249,
review by moors.
|
| |
|
|
|
|
| |
Reverted previous commit as I think it is wedging under -optimise. No
review.
|
| |
|
|
|
|
|
|
| |
Made some changes to fjbg so when people run into #1133 at least it will
tell them what method was the cause. The fact that ten files are touched
in this commit instead of one or two is a testament to the wonder of
checked exceptions. No review.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
ldc being used instead of bipush or sipush. The cheaper instructions
were only being used when the argument was typed as byte or short, but
an Int in the byte or short range can as easily use them. This also
saves an entry in the constant pool.
iconst_n not being used except on Ints. Same issue in reverse: pushing
short 3 on the stack is cheaper with iconst_3 than with sipush.
Example:
class A {
def f1: Int = 100 // was ldc #X, now bipush 100
def f2: Int = 5000 // was ldc #X, now sipush 5000
def f3: Byte = 2 // was bipush 2, now iconst_2
def f4: Short = 2 // was sipush 2, now iconst_2
}
Review by dragos.
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Restoring negative literal parsing behavior to what should be the least
surprising option. Thanks much to Johannes Rudolph for identifying the
bug in the bytecode generator which needed addressing for us to arrive
at proper -0.0 behavior, and for writing the majority of this patch.
A '-' followed immediately by either a number or a period should now
always be treated as a single numeric literal, which means the minus
binds more tightly than anything else. A specific example of how this
differs from 2.8 final is:
-5.+(10) == 5.0 // and not -15.0
The full range of potentially ambiguous parses involving prefix
operators, numbers, and dots is quite large and still needs to be
completely and clearly specified.
Closes #2378 and #3657, review by odersky, jrudolph.
|
| |
|
|
|
|
| |
Closes #3310 (very large Scala class is compiled to invalid classfile
because Scala signature can't fit into constant pool). Review by dragos.
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
Scala signature is generated as an annotation (that is accessible
through Java reflection).
- compiler generates all pickled Scala signatures as annotations to class files.
- compiler can read class files with signature as annotations or old-style signatures as attributes.
- Scalap has also been updated to new signatures (contributed by Ilya Sergey: thanks a lot).
- FJBG updated to allow entering constant pool strings as byte arrays.
- ByteCodecs decode method returns the length of the decoded array.
Review by ilyas. Already mostly reviewed by odersky.
|
| |
|
|
|
|
|
| |
fixing a long-standing bug in fjbg and recompiling fjbg.jar, which had
the side effect of revealing that the current fjbg jar had never been
recompiled with target 1.5, so now it's smaller and (I imagine) faster.
|
| | |
|
| |
|
|
|
|
| |
Code to support invoke-dynamic for structural types. Not yet complete,
built around the JSR 292 spec of December 2008.
|
| | |
|
| |
|
|
|
| |
Added experimental invoke dynamic support (JSR292).
|
| |
|
|
|
|
| |
Inner classes follow their enclosing class to satisfy the Eclipse Java
compiler
|
| | |
|
| | |
|
| |
|
|
|
|
| |
Added patch to allow definition of native methods in Scala (contribution
from Stepan Koltsov)
|
| |
|
|
|
| |
Fixed many bugs in icodeReader and added basic blocks.
|
| |
|
|
|
| |
Updated jars to target java 1.4 and work on 1.4 VMs
|
| |
|
|
|
|
| |
Fixed bug #680 and #455 (duplicates) by carrying more typing context to
the backend for method calls.
|
| | |
|
| | |
|
| |
|
|
|
| |
Added support for emitting Java annotations into classfiles
|
| |
|
|
|
|
| |
Added fjbg.jar contents to the scala-compiler.jar so that the STARR
compiler accepts the new methods in fjbg to build locker.
|
| | |
|
|
|
Switching to the new build system and to the new build system. This is a
MAJOR commit, so be careful when updating.
|
