FishEye: history openjfx-compiler/trunk/src/share/classes/com/sun/tools/javafx/comp/JavafxToJava.java

4868 annotated / raw | Diffs: previous, other | Lines: 3557 ( +250, -336 )

Created: 2009-01-06 12:45:54 -0800 (31 hours ago) | Author: rfield | Changeset: 4868

Fix JFXC-2606 : Compiler error returing an if expr that has a throw in the else part
caused by bollixed-up translation state. fixed by the below reorg. test added

Fix JFXC-2610 : Regularize and localize state transitions in translation
the biggest mess remaining in translation was the handling of global state (Location vs Value, Expression vs Statement, and target type) which really should not be global state at all.
they are now passed as parameters, and the visitor methods retrieve them at the boundary.
considerable code deletion and simplification.

4855 annotated / raw | Diffs: previous, other | Lines: 3643 ( +58, -44 )

Created: 2009-01-03 22:24:18 -0800 (3 days ago) | Author: rfield | Changeset: 4855

Fix JFXC-2592 : KeyFrame/Value retranslation hack blocks elision
complete removal of retransformation hack.
unify Instanciation translation with new KeyFrame/Value translation.
remove final retransformation in Instanciation -- sequence upcast.

Note that above also fixes --
Fix JFXC-1087 : NPE in bind for
re-enable test and move to regression tests.

Also, clean-up in parser support.

4854 annotated / raw | Diffs: previous, other | Lines: 3629 ( +41, -39 )

Created: 2009-01-03 20:04:06 -0800 (4 days ago) | Author: rfield | Changeset: 4854

Progress JFXC-2592 : KeyFrame/Value retranslation hack blocks elision
fix JFXKeyFrameLiteral to KeyFrame translation

4853 annotated / raw | Diffs: previous, other | Lines: 3627 ( +66, -53 )

Created: 2009-01-03 19:18:06 -0800 (4 days ago) | Author: rfield | Changeset: 4853

Progress JFXC-2592 : KeyFrame/Value retranslation hack blocks elision
fix InterpolateValue to KeyValue translation

4852 annotated / raw | Diffs: previous, other | Lines: 3614 ( +129, -82 )

Created: 2009-01-03 17:40:21 -0800 (4 days ago) | Author: rfield | Changeset: 4852

Progress JFXC-2592 : KeyFrame/Value retranslation hack blocks elision
restructure InstanciateTranslator so that the new super-class can be used directly by KeyFrame/Value.

4821 annotated / raw | Diffs: previous, other | Lines: 3567 ( +25, -25 )

Created: 2008-12-20 01:00:32 -0800 (18 days ago) | Author: rfield | Changeset: 4821

failure count = 5 (1 bug fixed, 1 test fixed, 1 punt, 2 fixes uncovered)
and more, see below

Fixed JFXC-2559 : Disparate numeric types in bound explicit sequence cause crash
fix translation of bound explicit sequence
add regression test

Fixed JFXC-2551 : Number->Float: FAILED regress/jfxc848boundLiteral.fx
fixed by above, plus test using Double in order to get expected results

Progress on JFXC-2560 : Attribution does not propagate expected type in bound explicit sequence numeric conversion
singleton bound explicit numeric sequences -- attribution fixed, translation still needs fix (but not a regression, so save for later)
currently-failing test added

Fixed JFXC-2557 : Number->Float: FAILED operators/Operators3.fx
test should be Double, so changed to that.
output was thousands of characters per line and not comparable, changed to twenty numbers per line

Punt on JFXC-2552 : Number->Float: FAILED regress/jfxc915a.fx
This error is a missing loss of precision error on Long->Integer, so not a show-stopper
moved test to currently-failing

Close JFXC-2556 : Number->Float: FAILEDAccessModifiers/Derived1.fx
Close JFXC-2558 : Number->Float: FAILED should-fail/AccessModifiersTest.fx
no longer reproducible

Branch point for: franca

Tags: m8

4800 annotated / raw | Diffs: previous, other | Lines: 3567 ( +13, -4 )

Created: 2008-12-18 21:05:08 -0800 (19 days ago) | Author: rfield | Changeset: 4800

Fixed JFXC-2539 : Number->Float: range sequences crash for new numerics
test now passes, I'm now getting 24 failures

then fix a couple of .EXPECTED files:
Progress JFXC-2527 : Number->Float: regress/jfxc1779.fx fails at runtime - output does not match expected output
defaultseq.fx
jfxc1779.fx

4798 annotated / raw | Diffs: previous, other | Lines: 3558 ( +2, -2 )

Created: 2008-12-18 18:37:13 -0800 (20 days ago) | Author: rfield | Changeset: 4798

Fixed JFXC-2451 : assignops fall over on new numerics
operation type forByte and Short is Integer.

4796 annotated / raw | Diffs: previous, other | Lines: 3558 ( +6, -4 )

Created: 2008-12-18 17:15:51 -0800 (20 days ago) | Author: rfield | Changeset: 4796

Fixed JFXC-2515 : Number->Float: regress test jfxc2269.fx fails in translation "mul(float) in javafx.lang.Duration cannot be applied to (double)"
jfxc2269.fx now works

4779 annotated / raw | Diffs: previous, other | Lines: 3556 ( +1, -1 )

Created: 2008-12-17 22:25:24 -0800 (20 days ago) | Author: rfield | Changeset: 4779

Progress JFXC-2412 : Method resolution for numeric types
fixes the problem, however, causes jfxc915a.fx to fail (warnings about Long->Integer conversion are not reported).
so, the code in JavafxTypes has been commented out (the code below should be removed).
passing it back to Sundar.
test code in bug report.

4778 annotated / raw | Diffs: previous, other | Lines: 3556 ( +8, -10 )

Created: 2008-12-17 20:09:15 -0800 (20 days ago) | Author: rfield | Changeset: 4778

Fixed JFXC-2510 : Translation: conversion between bound numeric sequences
use new runtime conversion.
factor-out common code with unbound case.

Enable test for JFXC-2492 Runtime: conversion between bound numeric sequences
since it is now being used.

4777 annotated / raw | Diffs: previous, other | Lines: 3558 ( +2, -7 )

Created: 2008-12-17 17:34:44 -0800 (21 days ago) | Author: briangoetz | Changeset: 4777

JFXC-2494 -- make "lazy" parameter explicit on unary bind methods (those that take only another Location). Obviates need for isLazilyBound() on Location, as compiler is now passing lazy status all the time, so that is removed. Translation changes from Robert.

4772 annotated / raw | Diffs: previous, other | Lines: 3563 ( +7, -4 )

Created: 2008-12-17 00:11:23 -0800 (21 days ago) | Author: sundararajana | Changeset: 4772

Fixed JFXC-2443 - Get rid of all the assumptions in attribution about
the numeric types.

-- in "partial" commits for this fix, I had changed Number->Double in javafxdoc
code. Changed that to use "Number" depending on choice of Number type in
compiler invocation (-XDNumber=...)

Also, while cleaning up all "TODO:NUMERIC" stuff, I ended up adding translation
support for conversion between numeric sequences. Hence, this change fixes
JFXC-2442 as well.

Fixed JFXC-2442 - Runtime: conversions between numeric sequences.

4765 annotated / raw | Diffs: previous, other | Lines: 3560 ( +4, -2 )

Created: 2008-12-15 04:40:25 -0800 (23 days ago) | Author: mauriziocima | Changeset: 4765

Fixed JFXC-1331: untyped empty sequence is translated to Object
Added empty sequence symbol to JavafxSymtab; adjusted attribution and translation of empty sequence literals accordingly.

4762 annotated / raw | Diffs: previous, other | Lines: 3558 ( +4, -1 )

Created: 2008-12-14 21:27:20 -0800 (23 days ago) | Author: rfield | Changeset: 4762

Fixed JFXC-2466 : Lazy Binding: Operators -- runtime
implement lazy/eager switching on operators.
cross-product tests for lazily bound operator and operand type.

4760 annotated / raw | Diffs: previous, other | Lines: 3555 ( +1, -1 )

Created: 2008-12-14 00:06:16 -0800 (24 days ago) | Author: rfield | Changeset: 4760

Fixed JFXC-2477 : Correct conversion problems that show-up when Number=Float
fix conversion issue.
add cross-product tests for conversions between numeric types in a variety of situations.

4755 annotated / raw | Diffs: previous, other | Lines: 3555 ( +2, -2 )

Created: 2008-12-11 18:55:33 -0800 (27 days ago) | Author: rfield | Changeset: 4755

Fixed JFXC-2479 : Arithmetic operations on Duration by new numerics
When Number is 32-bit, operations on Duration break. This fixes so it works regardless.

4747 annotated / raw | Diffs: previous, other | Lines: 3555 ( +2, -6 )

Created: 2008-12-09 21:18:19 -0800 (28 days ago) | Author: rfield | Changeset: 4747

Fixed JFXC-2462 : Lazy binding: Syntax
Fixed JFXC-2463 : Lazy Binding: Propagate lazy flag to translation and into Locations
information propagated to runtime, and runtime still uses the information, but it no longer causes lazy behavior

4746 annotated / raw | Diffs: previous, other | Lines: 3559 ( +6, -2 )

Created: 2008-12-09 18:09:16 -0800 (29 days ago) | Author: bothner | Changeset: 4746

Merge changeset 4682 from perf-sandbox branch:
Get rid of SequenceBuilder. Instead, we use an ArraySequence directly.

4744 annotated / raw | Diffs: previous, other | Lines: 3555 ( +62, -6 )

Created: 2008-12-09 17:14:43 -0800 (29 days ago) | Author: bothner | Changeset: 4744

Undo changeset 4705, which reverted changeset 4664, bringing back:

Change ArraySequence to use a buffer-gap. Perhaps modification and
concatenation "in-place" if the 'shared' flag is false.
This implements JFXC-2336 "implement shared-unshared sequence optimization".

4717 annotated / raw | Diffs: previous, other | Lines: 3499 ( +10, -8 )

Created: 2008-12-02 20:52:59 -0800 (35 days ago) | Author: briangoetz | Changeset: 4717

JFXC-2345 -- implement key value target as a function instead of a binding. Patch from Chris. Had to fix a number of tests, including moving some tests to limbo because they want to modify the interpolator and reuse the timeline, which doesn't seem kosher.

This also broke JFXC-1087, so reopening that issue (moved test to currently-failing.)

Tags: trunk-checkpoint

4716 annotated / raw | Diffs: previous, other | Lines: 3497 ( +2, -1 )

Created: 2008-12-02 20:20:06 -0800 (35 days ago) | Author: rfield | Changeset: 4716

Fixed JFXC-2410 : Predefined type names for {Character, Byte, Short, Long , Double, Float}
these type names can be used to define vars, etc.
test added which tests usage and verifies type.

Clean-up JFXC-2439 : Common numeric type definitions
remove all internal uses of javafx_NumberType -- instead explicitly use javafx_DoubleType.

4712 annotated / raw | Diffs: previous, other | Lines: 3496 ( +5, -5 )

Created: 2008-12-02 15:03:22 -0800 (36 days ago) | Author: rfield | Changeset: 4712

Fixed JFXC-2433 : Eliminate class literals
Consistently use TypeInfo (instead of mixture of Class<T> and TypeInfo) for identifying type in sequences.
This makes the code more mobile friendly.
Runtime changes by Brian.
Compiler support by Robert
Existing sequence tests cover this space.

Fixed JFXC-1955 : Compiler error while creating a sequence of functions in bind-context
currently failing test moved to regress.

Fixed JFXC-2402 : declaring sequence of functions crashes compiler
new regress test

4708 annotated / raw | Diffs: previous, other | Lines: 3496 ( +3, -1 )

Created: 2008-12-01 16:42:35 -0800 (37 days ago) | Author: JimHolmlund | Changeset: 4708

Fixed JFXC-2347: Regression: NPE on dereference
Added a case for SEQUENCE_INDEXED to JavafxToJava.possiblyNull

4705 annotated / raw | Diffs: previous, other | Lines: 3494 ( +5, -61 )

Created: 2008-11-30 21:21:30 -0800 (37 days ago) | Author: briangoetz | Changeset: 4705

Back out changeset 4664

4664 annotated / raw | Diffs: previous, other | Lines: 3550 ( +61, -5 )

Created: 2008-11-19 16:28:58 -0800 (49 days ago) | Author: bothner | Changeset: 4664

Change ArraySequence to use a buffer-gap. Perhaps modification and
concatenation "in-place" if the 'shared' flag is false.
This implements JFXC-2336 "implement shared-unshared sequence optimization".

4623 annotated / raw | Diffs: previous, other | Lines: 3494 ( +28, -5 )

Created: 2008-11-12 03:36:22 -0800 (56 days ago) | Author: sundararajana | Changeset: 4623

Fixing JFXC-1976 - Correctly set initial size of SequenceBuilder when
generating sequence literals.

Testing: Manually verified using javap that the compiler generates
calls to (initial) size accepting constructors.

4566 annotated / raw | Diffs: previous, other | Lines: 3471 ( +10, -4 )

Created: 2008-11-05 15:44:41 -0800 (2 months 2 days ago) | Author: bothner | Changeset: 4566

Allow indexing of Java arrays.
This fixes JFXC-2349 "Compiler crash indexing float array
returned from Java code".

Branch point for: perf-sandbox

4564 annotated / raw | Diffs: previous, other | Lines: 3465 ( +1, -8 )

Created: 2008-11-05 14:28:50 -0800 (2 months 2 days ago) | Author: bothner | Changeset: 4564

If required type is an array, we should only convert sequence-to-array
if the expression type is a sequence type, not if it is a non-sequence.
Specifically, we should leave null and java.lang.Object as-is.
This fixes JFXC-2348 "Regression: type inference on null and []".

4504 annotated / raw | Diffs: previous, other | Lines: 3472 ( +1, -2 )

Created: 2008-10-25 09:08:20 -0700 (2 months 13 days ago) | Author: rfield | Changeset: 4504

Fixed JFXC-1868 : Non-runnable scripts shouldn't be runnable (shouldn't have run, javafx$run$ or main)
User will now receive an exceptioin if they attempt to execute a library script without a run function.

Fixed JFXC-2250 : Could not compile JavaFX launcher MIDlet
1868 fixes 2250 by side-stepping a presumed 1.5 javac bug by just not generating the methods causing javac to get confused.

4496 annotated / raw | Diffs: previous, other | Lines: 3473 ( +2, -1 )

Created: 2008-10-24 00:43:33 -0700 (2 months 14 days ago) | Author: michael_heinrichs | Changeset: 4496

Update copyright-headers (JFXC-2148).

4462 annotated / raw | Diffs: previous, other | Lines: 3472 ( +28, -7 )

Created: 2008-10-20 18:29:17 -0700 (2 months 18 days ago) | Author: rfield | Changeset: 4462

Fixed JFXC-2167 : need to fix translation of INT *= NUMBER
Fixed JFXC-2269 : *= and /= for Duration do not compile
Overly simplistic type handling in *= and /=.
Both with regression tests.

Also, currently-failing should-fail test for JFXC-1449 : Check bound attributes changing

4448 annotated / raw | Diffs: previous, other | Lines: 3451 ( +5, -4 )

Created: 2008-10-17 20:24:54 -0700 (2 months 21 days ago) | Author: rfield | Changeset: 4448

Fixed JFXC-1779 : Result of a bound function is not re-evaluated when a function dependency changes
Function variables were not being correctly implemented in bound contexts.
They now update trigger on change and guard against NPE.
New white box tests.
Submitted test moved to regress.

4432 annotated / raw | Diffs: previous, other | Lines: 3450 ( +80, -47 )

Created: 2008-10-16 17:43:22 -0700 (2 months 22 days ago) | Author: rfield | Changeset: 4432

Fixed JFXC-2200 : Compiler crash java.lang.ProcessBuilder constructor
Problem was that varargs constructors were not handled.
With regression test.
Also fix issues with passing to array parameters.

4423 annotated / raw | Diffs: previous, other | Lines: 3417 ( +10, -6 )

Created: 2008-10-16 11:50:11 -0700 (2 months 22 days ago) | Author: briangoetz | Changeset: 4423

JFXC-2076; move conversion methods out of Sequences, rename toArray to toDoubleArray.

4421 annotated / raw | Diffs: previous, other | Lines: 3413 ( +14, -7 )

Created: 2008-10-15 19:44:14 -0700 (2 months 23 days ago) | Author: rfield | Changeset: 4421

Fixed JFXC-2070 : Compilation of script fails if String sequence contains null some times
Handle literal nulls in explicit sequenc creation.

Fixed JFXC-2204 : On assignment, object literal init, or initial value, convert literal null to default value
Fixed JFXC-2205 : Literal null or an Object value of null when cast to String/Duration must be converted to default value
These represent partial functionality for JFXC-960 : Umbrella: String/Duration types should never have null value
With regression tests and fix to the regression test for JFXC-2000 : Inconsistent treatment of nulls in sequences

Add currently-failing test for JFXC-2217 : On assignment, object literal init, or initial value, convert (non-literal) nullable value

Repaired currently-failing test for JFXC-103 which was missing the '/fail'

4364 annotated / raw | Diffs: previous, other | Lines: 3406 ( +77, -76 )

Created: 2008-10-08 19:31:24 -0700 (2 months 30 days ago) | Author: rfield | Changeset: 4364

Fixed JFXC-2124 : Function call arguments must be evaluated exactly once, even if null check fails

4362 annotated / raw | Diffs: previous, other | Lines: 3405 ( +15, -15 )

Created: 2008-10-08 16:14:25 -0700 (2 months 30 days ago) | Author: bothner | Changeset: 4362

Replace Name.fromString(names, "foo") by names.fromString("foo").
Not only is the latter simpler, the former no longer works with
the current javac sources.

4353 annotated / raw | Diffs: previous, other | Lines: 3405 ( +0, -4 )

Created: 2008-10-08 09:26:10 -0700 (2 months 30 days ago) | Author: JimIdle | Changeset: 4353

Fixed JFXC-1715

Cleans up the AST apis and interfaces so that the Interpolate node, which is
no longer used, does not confuse external developers.

Ensures that the JavaFX tree scanner will visit the interpolation values and PathFinder
will therefore visit and resolve them.

4346 annotated / raw | Diffs: previous, other | Lines: 3409 ( +17, -20 )

Created: 2008-10-07 14:46:14 -0700 (3 months ago) | Author: bothner | Changeset: 4346

Generalize type-casts of primitives to work regardless of boxing
(i.e. whether source/target type is int *or* java.lang.Integer).

Fix casting of insert value or sequence constructor value,
converting to element or sequence tpe appropriatel.

This fixes JFXC-2047 "Compilation fails if we add a double value to integer sequence".

4335 annotated / raw | Diffs: previous, other | Lines: 3412 ( +3, -3 )

Created: 2008-10-06 19:24:31 -0700 (3 months 1 day ago) | Author: rfield | Changeset: 4335

Fixed JFXC-2146 : Regression: Type conversion problem on assignment to select by non-simple expression -- found in Assortis ElectroSimulator

4323 annotated / raw | Diffs: previous, other | Lines: 3412 ( +1, -0 )

Created: 2008-10-06 11:45:13 -0700 (3 months 1 day ago) | Author: sundararajana | Changeset: 4323

Fixing JFXC-1663 -Xjcov option cause exception in generate().

Test:

* compiled a simple FX program with -Xjcov and verified that the
compilation finishes normally.

* Verified that the compiler generated .class files have CharacterRangeTable
attribute (used by jcov) when using -Xjcov option. [used "jclasslib" to
view .class files]

Branch point for: m7-release tag:milestone-7

Tags: milestone-7

4308 annotated / raw | Diffs: previous, other | Lines: 3411 ( +47, -9 )

Created: 2008-10-04 23:00:02 -0700 (3 months 2 days ago) | Author: rfield | Changeset: 4308

Copied to: branches/m7-release/trunk/src/share/classes/com/sun/tools/javafx/comp/JavafxToJava.java 4321

Fixed JFXC-1902 : bind with inverse inside object literals doesn't sync up values
Fixed by adding generalizable substitution mechanism.
Currently-failing test moved to regress and enhanced with many more combinations.

Branch point for: m7-release tag:milestone-7

Tags: milestone-7

4304 annotated / raw | Diffs: previous, other | Lines: 3373 ( +41, -99 )

Created: 2008-10-04 00:02:16 -0700 (3 months 3 days ago) | Author: rfield | Changeset: 4304

Misc, in process, fixes and clean-ups

4297 annotated / raw | Diffs: previous, other | Lines: 3431 ( +0, -2 )

Created: 2008-10-03 18:51:35 -0700 (3 months 4 days ago) | Author: JimIdle | Changeset: 4297

Fixed JFXC-1766

This change removes fx types that are no longer relevant, at the request of the Netbeans
IDE team, as it is unclear to them what things they need to support when there is old
stuff in there. Because of the proximity of the cut-off I did not remove types and classes
that were widely used in the source code.

This change removes checks for things that are now removed from the code.

This change also finalizes the postion of the lexer/parser for release, in terms of
how it reports on keywords and operators that are no onger supported. In general, warings
are now errors, directing the user at the offical way to do things. The exception is attribute,
which is still a warning because the preview supported attribute and this release would
be a pain if it absolutely forced people to change all their code for that.

Some error messages were elaborated upon, and regression and should-fail tests were updated
to reflect the error message changes.

4295 annotated / raw | Diffs: previous, other | Lines: 3433 ( +2, -2 )

Created: 2008-10-03 15:59:08 -0700 (3 months 4 days ago) | Author: bothner | Changeset: 4295

Coerce slice index expressions to int when translating.
This fixes JFXC-2078 "Error and warning for use of a Number in a
sequence slice expression".

4260 annotated / raw | Diffs: previous, other | Lines: 3433 ( +2, -0 )

Created: 2008-10-01 16:18:56 -0700 (3 months 6 days ago) | Author: rfield | Changeset: 4260

Fixed JFXC-2114 : NPE when running RT_423Test -- should be caught
Simple bug in null checking.
With minimal-form test and mimic test.

4257 annotated / raw | Diffs: previous, other | Lines: 3431 ( +2, -19 )

Created: 2008-10-01 14:31:39 -0700 (3 months 6 days ago) | Author: rfield | Changeset: 4257

Fixed JFXC-2110 : Failed to compile with generics error on simple script
Deferred Location requirement determination was underlying cause, move all this to JavafxVarUsageAnalsysis.
This, in turn, required that function values don't permeate bind.
Regression test based on what Chris sent.

4247 annotated / raw | Diffs: previous, other | Lines: 3448 ( +1, -7 )

Created: 2008-09-30 22:27:03 -0700 (3 months 6 days ago) | Author: bothner | Changeset: 4247

Optimize, inline, and remove isJavaLangObjectType method.

4242 annotated / raw | Diffs: previous, other | Lines: 3454 ( +1, -2 )

Created: 2008-09-30 19:34:20 -0700 (3 months 7 days ago) | Author: rfield | Changeset: 4242

Fixed JFXC-2045 : Add isInitialize usage to VARUSE and check it
Detect which variables are used with isInitialized, and leave them Locations.

4238 annotated / raw | Diffs: previous, other | Lines: 3455 ( +1, -1 )

Created: 2008-09-30 12:37:44 -0700 (3 months 7 days ago) | Author: rfield | Changeset: 4238

Fixed but switch not thrown JFXC-2025 : Elide unbound script-private member vars
With these changes, passes all tests.
Switch not thrown until checked against GUI et. al.
Regression test cannot be committed until switch thrown.
Also, the jfxc1345.fx currently-failing test will need to be changed since the non-Location does not tickle the bug.
Changes --
Reflection: determination of member vars.
Increment of select with setters.

4235 annotated / raw | Diffs: previous, other | Lines: 3455 ( +10, -5 )

Created: 2008-09-30 09:10:54 -0700 (3 months 7 days ago) | Author: rfield | Changeset: 4235

Progress JFXC-2025 : Elide unbound script-private member vars
Handling of Duration assign ops.

4232 annotated / raw | Diffs: previous, other | Lines: 3450 ( +4, -3 )

Created: 2008-09-30 01:46:40 -0700 (3 months 7 days ago) | Author: rfield | Changeset: 4232

Progress JFXC-2025 : Elide unbound script-private member vars
Proper inter-script handling of setter.
Handling of special Duration TYPE_INFO variable.
Assignment to raw statics.

4216 annotated / raw | Diffs: previous, other | Lines: 3449 ( +79, -220 )

Created: 2008-09-28 23:56:53 -0700 (3 months 8 days ago) | Author: rfield | Changeset: 4216

Fixed JFXC-2068 : Use NullChecker for select
Introduce SelectTranslator, a subclass of NullChecker, to do all select processing cleanly and with consistent null checks.
Fixes numerous problems with select esp for select of a function.
With regression test.

Fixed JFXC-2067 : Use NullChecker for function invocation
Use the NullChecker to fix function call problems and clean the code as needed for elision.

These are the two issues in the umbrella (I wasn't able to fix one without fixing the other). So, the umbrella is fixed...
Fixed JFXC-2066 : Standardized NullChecker should be used universally
As a result, the old null checking crud has been deleted.

4212 annotated / raw | Diffs: previous, other | Lines: 3590 ( +127, -125 )

Created: 2008-09-28 01:59:48 -0700 (3 months 9 days ago) | Author: rfield | Changeset: 4212

Fixed JFXC-2083 : No null checks when assigning to a Java field
Plus many other missing null checks.
Preserve side-effect of RHS when LHS fails null check.
Richer infrastructure in null checker.
And other work needed for elision.

4211 annotated / raw | Diffs: previous, other | Lines: 3588 ( +130, -38 )

Created: 2008-09-27 22:31:27 -0700 (3 months 9 days ago) | Author: rfield | Changeset: 4211

Progress JFXC-2083 : No null checks when assigning to a Java field
(work needed for elision)
Centralize assign processing (==, ++, +=, etc) and null checks into AssignTranslator.

4208 annotated / raw | Diffs: previous, other | Lines: 3496 ( +8, -14 )

Created: 2008-09-27 14:45:51 -0700 (3 months 10 days ago) | Author: rfield | Changeset: 4208

Progress JFXC-2067 : Use NullChecker for function invocation
Convert NullChecker to a Translator NullCheckTranslator

4205 annotated / raw | Diffs: previous, other | Lines: 3502 ( +7, -4 )

Created: 2008-09-27 13:41:55 -0700 (3 months 10 days ago) | Author: rfield | Changeset: 4205

Progress JFXC-2067 : Use NullChecker for function invocation

4199 annotated / raw | Diffs: previous, other | Lines: 3499 ( +18, -24 )

Created: 2008-09-27 11:41:22 -0700 (3 months 10 days ago) | Author: rfield | Changeset: 4199

Progress JFXC-2025 : Elide unbound script-private member vars
Fixes to null checker.
Other misc fixes.

4197 annotated / raw | Diffs: previous, other | Lines: 3505 ( +12, -8 )

Created: 2008-09-27 00:44:28 -0700 (3 months 10 days ago) | Author: rfield | Changeset: 4197

Progress JFXC-2067 : Use NullChecker for function invocation
As it turns out, it appears the consistent functioning of null check is needed by Location-elision.
All work in commented-out code.

4196 annotated / raw | Diffs: previous, other | Lines: 3501 ( +1, -1 )

Created: 2008-09-26 16:19:09 -0700 (3 months 11 days ago) | Author: rfield | Changeset: 4196

Fixed JFXC-2077 : Compiler confused with assigning member variables on Java classes
fixed while listening to a very interesting talk on the Parrot VM

4192 annotated / raw | Diffs: previous, other | Lines: 3501 ( +8, -3 )

Created: 2008-09-26 10:27:22 -0700 (3 months 11 days ago) | Author: rfield | Changeset: 4192

Progress JFXC-2025 : Elide unbound script-private member vars
Translate class references when in JFXIdent.
Mostly a sync, as I'm switching to laptop.

4190 annotated / raw | Diffs: previous, other | Lines: 3496 ( +3, -2 )

Created: 2008-09-26 00:54:13 -0700 (3 months 11 days ago) | Author: rfield | Changeset: 4190

Progress JFXC-2025 : Elide unbound script-private member vars
Initialization of static non-Locations.
Null checking of non mutable expressions.

4189 annotated / raw | Diffs: previous, other | Lines: 3495 ( +73, -48 )

Created: 2008-09-25 21:08:31 -0700 (3 months 11 days ago) | Author: rfield | Changeset: 4189

Fixed JFXC-2043 : Assignment (=, +=,++,...) and applyDefaults must use setter methods
Assignment-like operations now support non-Location setter-based instance variables.
Definitional assignment translation restructured to support non-Location setter-based instance variables.

4184 annotated / raw | Diffs: previous, other | Lines: 3470 ( +221, -0 )

Created: 2008-09-24 22:11:43 -0700 (3 months 12 days ago) | Author: rfield | Changeset: 4184

Fixed JFXC-2042 : Centralized null check facility needed by elision
General mechanism implemented.
Not yet used.

Commented-out partial JFXC-2067 : Use NullChecker for function invocation
NullChecker client for function invocation used in the development of the NullChecker.
Code not quite finished and too great a change for this late in the release.

As the above will address JFXC-2056 : Calling a variable of function type throws NPE if it is null
This represents a partial commented-out fix fo rthis bug as well.
A currently-failing bug has been added for this.

4173 annotated / raw | Diffs: previous, other | Lines: 3249 ( +4, -2 )

Created: 2008-09-23 03:00:53 -0700 (3 months 14 days ago) | Author: michael_heinrichs | Changeset: 4173

Extended Sequence.toArray() with two parameters: sourceOffset and length. They can be used to convert only a slice of a sequence to an array. This is used to improve SubSequence.toArray() and CompositeSequence.toArray(), because only the needed parts of the underlying sequences are requested (JFXC-1966).

4170 annotated / raw | Diffs: previous, other | Lines: 3247 ( +25, -3 )

Created: 2008-09-22 11:30:22 -0700 (3 months 15 days ago) | Author: rfield | Changeset: 4170

Progress JFXC-2025 : Elide unbound script-private member vars
Misc changes. I need to switch to the null check aspect of the problem now.

4162 annotated / raw | Diffs: previous, other | Lines: 3225 ( +48, -73 )

Created: 2008-09-20 19:20:33 -0700 (3 months 17 days ago) | Author: rfield | Changeset: 4162

Progress JFXC-2025 : Elide unbound script-private member vars
Factor out variable sets in translation: sequence indexed, Location. non-Location.

4161 annotated / raw | Diffs: previous, other | Lines: 3250 ( +13, -13 )

Created: 2008-09-20 18:07:05 -0700 (3 months 17 days ago) | Author: rfield | Changeset: 4161

Progress JFXC-2025 : Elide unbound script-private member vars
Do some normalization of assign/assign-op processing.
As a side-effect fix a, so far, unreported type conversion problem in assign-op (eg +=)

4159 annotated / raw | Diffs: previous, other | Lines: 3250 ( +39, -32 )

Created: 2008-09-20 15:37:08 -0700 (3 months 17 days ago) | Author: rfield | Changeset: 4159

Progress JFXC-2025 : Elide unbound script-private member vars
Change initialization builder to handle member vars which aren't Locations.
Initial work on access to non-Location member vars.

4157 annotated / raw | Diffs: previous, other | Lines: 3243 ( +2, -2 )

Created: 2008-09-19 17:40:06 -0700 (3 months 18 days ago) | Author: rfield | Changeset: 4157

Progress JFXC-2025 : Elide unbound script-private member vars
Begin changing the logic of morphing.

4153 annotated / raw | Diffs: previous, other | Lines: 3243 ( +20, -29 )

Created: 2008-09-19 13:46:49 -0700 (3 months 18 days ago) | Author: rfield | Changeset: 4153

Progress JFXC-2025 : Elide unbound script-private member vars
Align morph and flag state in local var processing.

4148 annotated / raw | Diffs: previous, other | Lines: 3252 ( +23, -23 )

Created: 2008-09-19 10:09:28 -0700 (3 months 18 days ago) | Author: rfield | Changeset: 4148

Progress JFXC-2025 : Elide unbound script-private member vars
In preparation for complex morphing logic, clean-up morphing computation and access.
Gut VarMorphInfo.

4141 annotated / raw | Diffs: previous, other | Lines: 3252 ( +0, -6 )

Created: 2008-09-18 22:27:03 -0700 (3 months 18 days ago) | Author: rfield | Changeset: 4141

Fixed JFXC-2021 : Generate and store detailed variable usage info needed by Location elision
Rewrote JavafxVarUsageAnalysis to provide detailed usage information in the flag bits of the symbol.
Changed VarMorphInfo to use these bits instead of its previous limited information, but morph under (roughly) the same conditions as previously (changing that is next).
Remove incorrect ASSIGNED_TO compution from the attribution phase (JavafxVarUsageAnalysis now covers this).
Clean JavafxToJava to previous after-the-fact setting of morphability, all of which is now done up-front in JavafxVarUsageAnalysis.

Also, fix incorrect error bailout (was going into JavafxVarUsageAnalysis even with error).

4125 annotated / raw | Diffs: previous, other | Lines: 3258 ( +20, -10 )

Created: 2008-09-18 05:41:32 -0700 (3 months 19 days ago) | Author: mauriziocima | Changeset: 4125

Fixed JFXC-1743 : Problem with "inconvertible types"
Fixed JFXC-2004 : Ambiguous assignment
Fixed JFXC-2012 : compilation fails if sequence contains an embedded array returned from java
All those fixes required some adjustments both to the attribution phase and to the translation phase in order to achieve better interoperability with Java primitive types.

4119 annotated / raw | Diffs: previous, other | Lines: 3248 ( +1, -1 )

Created: 2008-09-17 19:50:54 -0700 (3 months 20 days ago) | Author: rfield | Changeset: 4119

Progress JFXC-2021 : Generate and store detailed variable usage info needed by Location elision
Set-up flags.

4113 annotated / raw | Diffs: previous, other | Lines: 3248 ( +31, -17 )

Created: 2008-09-17 14:53:18 -0700 (3 months 20 days ago) | Author: rfield | Changeset: 4113

Fixed JFXC-2017 : Cannot assgin an Integer sequence to a Number sequence
With regression test.

Fixed JFXC-1780 : setAsInt(int) in IntVariable cannot be applied to (long)
Which caused the JFXC-915 tests to fail, hence the movement of these tests.

4100 annotated / raw | Diffs: previous, other | Lines: 3234 ( +26, -23 )

Created: 2008-09-16 20:44:26 -0700 (3 months 21 days ago) | Author: rfield | Changeset: 4100

Partial JFXC-1982 : Eliminate reification of sequence element types
Use new TypeInfo instead of Class in the construction of sequences.
TypeInfo is then used to create default values and empty sequences.
Regression test added that checks defaults.

Runtime changes by Brian.
Compiler changes by Robert.

Also regression test for 849 changed to not depend on the implementing class name.

4082 annotated / raw | Diffs: previous, other | Lines: 3231 ( +69, -112 )

Created: 2008-09-15 10:18:44 -0700 (3 months 22 days ago) | Author: briangoetz | Changeset: 4082

Back out changes in r4073 (array syntax). To avoid merge conflicts, also had to back out r4075, which will reviewed and possibly reapplied.

4075 annotated / raw | Diffs: previous, other | Lines: 3274 ( +6, -10 )

Created: 2008-09-14 20:38:02 -0700 (3 months 23 days ago) | Author: chrisoliver | Changeset: 4075

avoid location wrapping for array element assignments

4073 annotated / raw | Diffs: previous, other | Lines: 3278 ( +115, -68 )

Created: 2008-09-14 18:48:43 -0700 (3 months 23 days ago) | Author: chrisoliver | Changeset: 4073

Implemented the majority of support for java arrays; Declarator syntax is the element type surrounded by a pair of brackets for each dimension, e.g var s:[String] declares an array of String. Interface is the same as java: elements may be accessed and assigned via []. A length property returns the length. Allocation is through a pseudo java constructor call, e.g var s = new [String](100), which is equivalent to java: String[] s = new String[100]. The divergence from java array declarator syntax is unavoidable due to the current syntax for sequences. Fwiw, EcmaScript 4 has a similar syntax for arrays as that implemented here. Implicit coercions from arrays to sequences and vice-versa have been preserved for now for backward compatiblity. However, these must be removed for 2 reasons: 1) Very error-prone: huge performance hits could easily occur due to conversion unnoticed by the user. 2) Incompatible with the type system: it should be possible to create a sequence of arrays, which would be ambiguous with the implicit coercion of an array to a sequence. TBD, a proper error report is lacking for an illegal argument in the constructor invocation.

4071 annotated / raw | Diffs: previous, other | Lines: 3231 ( +1, -1 )

Created: 2008-09-14 17:12:17 -0700 (3 months 23 days ago) | Author: JimIdle | Changeset: 4071

Progress JFXC-1117

Fix broken encapsulation in Javafx tree walker.
Raise the getErrorNodes() method to base class JFXTree so that any
node can masqerade as an ERRONEOUS type without the need to cast
trees to Erroneous.

4067 annotated / raw | Diffs: previous, other | Lines: 3231 ( +0, -5 )

Created: 2008-09-14 13:18:12 -0700 (3 months 23 days ago) | Author: JimIdle | Changeset: 4067

Progress JFXC-1117

All remaining parser rules now accumulate Erroneous AST nodes and recover from
syntax nodes.

Remove MissingTree interface as IDE guys don't want it.

Rename MissingXXX classes to ErroneousXXX.

Fix up AST walking classes because of removal of MissingTree based AST nodes.

4046 annotated / raw | Diffs: previous, other | Lines: 3236 ( +7, -7 )

Created: 2008-09-12 06:28:05 -0700 (3 months 25 days ago) | Author: mauriziocima | Changeset: 4046

Fixed JFXC-1730 : Compilation error in doing Duration += n

4033 annotated / raw | Diffs: previous, other | Lines: 3236 ( +6, -1 )

Created: 2008-09-11 09:24:53 -0700 (3 months 26 days ago) | Author: mauriziocima | Changeset: 4033

Fixed JFXC-1681 : Return statement inconsistency
Fixed JFXC-1299 : Cannot infer "Void" return type in function with "Void" "return" statement
The two fixes are deeply intertwined; they both have to do with failures in return type inference that is corrected by this fix

4028 annotated / raw | Diffs: previous, other | Lines: 3231 ( +2, -2 )

Created: 2008-09-10 21:07:35 -0700 (3 months 26 days ago) | Author: rfield | Changeset: 4028

More JFXC-1957 : Track and print optimization statistics
Bound and unbound var/def -- Location vs direct

4015 annotated / raw | Diffs:

FishEye

View detailed chart Line History

View Mode

Show Arbitrary Diff

Watches and RSS

History

latest revision trunk