summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* java 5 supportacolyer2005-01-1745-159/+882
|
* Fix for AjcPre10xTests.test016 if run standalone. (if run standalone no-one ↵aclement2005-01-141-0/+4
| | | | was setting the world correctly for certain types and you'd get an NPE).
* Fix for "staticinitialization(@Blah *)"aclement2005-01-141-0/+2
|
* Fixed 3 bugs reported in bug#82340, including correct use of visibility ↵mkersten2005-01-137-104/+184
| | | | modifiers.
* Updated to support running on JDK 1.5, and fixed related bug#82218mkersten2005-01-136-26/+78
|
* Updated to include source location offset.mkersten2005-01-131-3/+6
|
* Fixes for 78021, 79554 - both to do with us breaking the exception table for ↵aclement2005-01-1112-41/+189
| | | | a method on weaving *if* finally blocks are involved.
* Fix for Bug 82134: AspectJ 5 M2 should implement backwards compatibility for ↵aclement2005-01-0770-151/+316
| | | | binary aspect form
* rebuilt with 2.0 weaver version attribute (aj1.2.1 had version 1.0)aclement2005-01-0714-0/+0
|
* All jars rebuilt with rev'd weaver version number 2.0 that we will use for ↵aclement2005-01-0719-0/+0
| | | | AspectJ5 (AJ1.2.1 had version 1.0)
* Testcode for migration testingaclement2005-01-077-0/+73
|
* Removed warning that no longer comes out for this test. We were putting out ↵aclement2005-01-071-1/+1
| | | | warnings about changes to serialveruid even if a decp implements was going to fail - we no longer put out this misleading warning.
* test was failing because a message was coming out twice - once when applying ↵aclement2005-01-062-2/+2
| | | | decp at source time, once when applying it at weave time. Now removed the reporting at source time.
* Removing 'run' calls from the testcases for now ... to get a build through !aclement2005-01-061-5/+5
|
* BWD: One extra check.aclement2005-01-051-1/+2
|
* BWD: Improved source loc stuff and helper methodsaclement2005-01-051-0/+20
|
* Changed order processing for weaving to match the recent changes in source ↵aclement2005-01-051-17/+68
| | | | order processing. Before weaving decp on type A, we look at supertypes and superinterfaces of A.
* BWD: Main part of the implementation, checks all the rules before allowing ↵aclement2005-01-051-11/+292
| | | | the DECP to proceed.
* BWD: improved source loc creationaclement2005-01-051-0/+15
|
* BWD: Improvements in source location handling allow us to now include a line ↵aclement2005-01-051-1/+1
| | | | number here.
* BWD: New error checking catches one more error in this testcase.aclement2005-01-051-0/+1
|
* Recent changes seem to have made some messages come out in a different order !aclement2005-01-052-6/+8
|
* BWD: Updated - now type processing order is sensible for applying decp, we ↵aclement2005-01-051-5/+9
| | | | don't get an error here.
* BWD: test casesaclement2005-01-052-0/+495
|
* BWD: Use new API on sourcetypebinding to remember original type hierarchy ↵aclement2005-01-051-12/+15
| | | | and don't put out weaving messages for decp.
* Binary Weaving Decp: Two changes: In SourceTypeBinding we can now remember ↵aclement2005-01-052-0/+0
| | | | the type hierarchy before we munge it up. In ClassFile we generate code for the original hierarchy rather than any munged variant. (this gets fixed up later when decp is done at weave time)
* test cases and fix for Bugzilla Bug 81863acolyer2005-01-0511-26/+56
| | | | Annotation matching using within() PCD doesn't appear to be working.
* Binary Weaving Decp: test codeaclement2005-01-0558-0/+362
|
* extra test resources for within(@Foo *) testsacolyer2005-01-052-0/+41
|
* Binary Weaving Decp: helper methods and bug fix for unnecessary suid warning.aclement2005-01-051-1/+10
|
* Binary Weaving Decp: another helper method..aclement2005-01-051-0/+51
|
* switch from string comparison to a more robust RPN format for testing ↵acolyer2005-01-051-8/+88
| | | | pointcut expression equality
* temporarily taken out a periodically misbehaving test caseacolyer2005-01-051-1/+2
|
* organized imports.aclement2005-01-051-6/+5
|
* Very very first simple helper bits of binary weaving decpaclement2005-01-042-0/+22
|
* Fix for Bug 82062: WeaveMessage should provide more informationaclement2005-01-045-11/+58
| | | | (Bug and fix from Martin Lippert)
* organized imports.aclement2005-01-045-21/+40
|
* Hopefully fixes disappearance of decsofts in AJDT. The addition of the ↵aclement2004-12-202-4/+4
| | | | offset to the source location caused problems because a declare soft was being assigned 2 diff offsets in different parts of the weaver. First source location was the offset of the 'declare', second location was the offset for the declares pointcut expression. Now resolved so offset is always position of the 'declare'.
* Fix for Bug 80916: In some cases the structure model doesn't contain the ↵aclement2004-12-201-0/+3
| | | | "matches declare" relationship
* incorporate tim peierls suggestion from mailing listacolyer2004-12-171-20/+11
|
* forgot to check in this updated java 5 jar for test casesacolyer2004-12-161-0/+0
|
* more ambiguous bindings testingacolyer2004-12-164-0/+121
|
* disambiguating bindingsacolyer2004-12-1610-1/+151
|
* it is now safe to short-circuit in and/or pc matching (all side-effects removed)acolyer2004-12-164-17/+20
|
* payback time for all the "kinding" work - only ever generate and matchacolyer2004-12-1640-334/+278
| | | | against shadows that we know might be matched.
* just tidying up a few unused importsacolyer2004-12-157-12/+0
|
* partial fix for bug 61568.acolyer2004-12-1550-156/+1794
| | | | | | | | | | | | | | | | | These changes are both (a) a performance optimization, and (b) an improvement on the binding across || rules that we implemented in 1.2.1. Instead of saying the the first binding in a left-to-right traversal of the pointcuts DNF is the one that you get (too much to ask users to do DNF rewriting in their heads), this version implements the rule that every || branch in the DNF must bind all formals, and if two || branches can have any join points in common (can match join points of the same kind), then both must bind all formals identically. So it allows things like execution(* *(..)) && args(x) || call(* *(..)) && this(x) which previously we used to forbid. But primarily it turned out to be a performance optimization.
* Support for 'offset' in source locations - enabling AJDT improvements. Not ↵aclement2004-12-1011-17/+63
| | | | perfect - really the whole ISourceLocation thing needs sorting out ...
* Updates for AJDK1.5.0aclement2004-12-102-2/+2
|
* Autoboxing and decp restrictionsaclement2004-12-1015-16/+248
|
generated by cgit v1.2.3 (git 2.39.1) at 2025-02-05 21:17:49 +0000