My favorites | Sign in
Logo
          
Details: Show all Hide all

Last 7 days

  • Dec 24, 2009
    r246 (1. Change entry-point and atomic-method finders to return on...) committed by arie.zilberstein   -   1. Change entry-point and atomic-method finders to return only application classes. Library classes are irrelevant here. 2. In interprocedural analysis, include the set of entry points in the set of reachable methods. 3.
    1. Change entry-point and atomic-method finders to return only application classes. Library classes are irrelevant here. 2. In interprocedural analysis, include the set of entry points in the set of reachable methods. 3.
  • Dec 24, 2009
    r245 (1. Don't tag library classes - they aren't transformed so no...) committed by arie.zilberstein   -   1. Don't tag library classes - they aren't transformed so no need to optimize them. (Information about them should, however, be collected, because they can affect application classes). 2. Library classes should not contribute any new points-to analysis nodes. They can only change existing nodes. (Why? If a node is not accessible from an application class, we won't need to optimize anything about it). 3. Analyze all methods, library + application. Ensure soundness.
    1. Don't tag library classes - they aren't transformed so no need to optimize them. (Information about them should, however, be collected, because they can affect application classes). 2. Library classes should not contribute any new points-to analysis nodes. They can only change existing nodes. (Why? If a node is not accessible from an application class, we won't need to optimize anything about it). 3. Analyze all methods, library + application. Ensure soundness.
  • Dec 24, 2009
    r244 (1. Catch up with soot 2. Let JarUtils extract multiple jars ...) committed by arie.zilberstein   -   1. Catch up with soot 2. Let JarUtils extract multiple jars (semicolon-separated) and improve its performance by working in chunks instead of in single bytes
    1. Catch up with soot 2. Let JarUtils extract multiple jars (semicolon-separated) and improve its performance by working in chunks instead of in single bytes

Last 30 days

  • Dec 18, 2009
    r243 (Add a launcher for offline instrument) committed by gkorland   -   Add a launcher for offline instrument
    Add a launcher for offline instrument
  • Dec 18, 2009
    r242 (Issue 2 Add a way to instrument more than one file at the ti...) committed by gkorland   -   Issue 2 Add a way to instrument more than one file at the time.
    Issue 2 Add a way to instrument more than one file at the time.
  • Dec 18, 2009
    r241 (Add launcher ) committed by gkorland   -   Add launcher
    Add launcher
  • Dec 17, 2009
    r240 (Allow AdviceBookkeeper to work even if all optimizations are...) committed by arie.zilberstein   -   Allow AdviceBookkeeper to work even if all optimizations are disabled
    Allow AdviceBookkeeper to work even if all optimizations are disabled
  • Dec 16, 2009
    r239 (Fix compilation problem) committed by arie.zilberstein   -   Fix compilation problem
    Fix compilation problem
  • Dec 16, 2009
    r238 (1. Merge distinct pairs instead of consecutive pairs. 2. Com...) committed by arie.zilberstein   -   1. Merge distinct pairs instead of consecutive pairs. 2. Comment the skipped method action for FieldActivity analysis.
    1. Merge distinct pairs instead of consecutive pairs. 2. Comment the skipped method action for FieldActivity analysis.
  • Dec 14, 2009
    r237 (Skip "SKIP" into SKIP_IMMUTABLE, SKIP_NEW_IN_CTOR, and SKIP_...) committed by arie.zilberstein   -   Skip "SKIP" into SKIP_IMMUTABLE, SKIP_NEW_IN_CTOR, and SKIP_NEW_LOCAL to allow collecting statistics.
    Skip "SKIP" into SKIP_IMMUTABLE, SKIP_NEW_IN_CTOR, and SKIP_NEW_LOCAL to allow collecting statistics.
  • Dec 14, 2009
    r236 (1. Drop track String objects in the Points-To Analysis. 2. A...) committed by arie.zilberstein   -   1. Drop track String objects in the Points-To Analysis. 2. Add a StmtsDatabase to map UniqueCodePoints to Stmts. This will make understanding the outputs much easier. 3. Bugfix for methods which do not access this.
    1. Drop track String objects in the Points-To Analysis. 2. Add a StmtsDatabase to map UniqueCodePoints to Stmts. This will make understanding the outputs much easier. 3. Bugfix for methods which do not access this.
  • Dec 06, 2009
    r235 (Keep track and report how many optimizations were enabled.) committed by arie.zilberstein   -   Keep track and report how many optimizations were enabled.
    Keep track and report how many optimizations were enabled.
  • Dec 06, 2009
    r234 (Don't choke on rt.jar) committed by arie.zilberstein   -   Don't choke on rt.jar
    Don't choke on rt.jar
  • Dec 06, 2009
    r233 (Add an interface to mark all code point databases.) committed by arie.zilberstein   -   Add an interface to mark all code point databases.
    Add an interface to mark all code point databases.
  • Dec 01, 2009
    r232 (1. Add a PureMethod analysis to discover "pure" methods: met...) committed by arie.zilberstein   -   1. Add a PureMethod analysis to discover "pure" methods: methods which do not need to be transactified since they don't access any field or array and they don't invoke impure methods. 2. Integrate with LastFieldActivity analysis: if the method invoked is pure, then the LFA property still holds.
    1. Add a PureMethod analysis to discover "pure" methods: methods which do not need to be transactified since they don't access any field or array and they don't invoke impure methods. 2. Integrate with LastFieldActivity analysis: if the method invoked is pure, then the LFA property still holds.
  • Nov 30, 2009
    r231 (1. Remove dead code (old intraproc analysis was replaced by ...) committed by arie.zilberstein   -   1. Remove dead code (old intraproc analysis was replaced by an interproc one) 2. Fix bug with NullPointerException on a statement which wasn't analyzed. It wasn't analyzed because the code was unreachable.
    1. Remove dead code (old intraproc analysis was replaced by an interproc one) 2. Fix bug with NullPointerException on a statement which wasn't analyzed. It wasn't analyzed because the code was unreachable.

Earlier this year

  • Nov 28, 2009
    r230 (1. Implement LastFieldActivity optimization by inserting an ...) committed by arie.zilberstein   -   1. Implement LastFieldActivity optimization by inserting an early commit. Make sure the real commit will play along with it. 2. Make the advice a bit field instead of a single value. 3. Fix for case of Jimple stmts mapping to multiple bytecode offsets.
    1. Implement LastFieldActivity optimization by inserting an early commit. Make sure the real commit will play along with it. 2. Make the advice a bit field instead of a single value. 3. Fix for case of Jimple stmts mapping to multiple bytecode offsets.
  • Nov 28, 2009
    r229 (Catch up with trunk) committed by arie.zilberstein   -   Catch up with trunk
    Catch up with trunk
  • Nov 23, 2009
    r228 (Fix 2 bugs in NewLocals: 1. Properly merge the valueToValueN...) committed by arie.zilberstein   -   Fix 2 bugs in NewLocals: 1. Properly merge the valueToValueNumber map in flowsets. 2. When creating a new leaky object, don't assign DefinitelyNew: assign SuspectedNotNew.
    Fix 2 bugs in NewLocals: 1. Properly merge the valueToValueNumber map in flowsets. 2. When creating a new leaky object, don't assign DefinitelyNew: assign SuspectedNotNew.
  • Nov 14, 2009
    r227 (Fix test for JDK1.6) committed by gkorland   -   Fix test for JDK1.6
    Fix test for JDK1.6
  • Nov 14, 2009
    r226 (Tag 1.3) committed by gkorland   -   Tag 1.3
    Tag 1.3
  • Nov 14, 2009
    r225 (Integrate with the field activity analysis. If a field is re...) committed by arie.zilberstein   -   Integrate with the field activity analysis. If a field is readonly, skip checking if it appears in the writeset.
    Integrate with the field activity analysis. If a field is readonly, skip checking if it appears in the writeset.
  • Nov 14, 2009
    issue 42 (Provide one external handler to listen to the course of the ...) reported by gkorland   -   we define a Method class that hold all informations we need on origin method. I added getAnnotations() method because i want to have the flexibility to allow other annotations that the handler can use (like @Persist) class Method { /* return method name */ public String getName(); /* return method parameters values or an empty array if none */ public Object[] getParameters(); /* return run time annotations or an empty array */ public Annotation[] getAnnotations(); } we define a TransactionEvent class class TransactionEvent { public static enum Type { INIT, COMMIT, ROLLBACK, RETRY, /* HOOK to allow, just before commiting, an external agent to do something that if throws an exception will rollback the transaction. */ HOOK }; public Type getType(); public Method getOriginMethod(); /* return the origin object (a class if method call was static) */ public Object getOriginObject(); } we define a TransactionEventHandler interface interface TransactionEventHandler { /* called after event in Deuce */ void handle(TransactionEvent event); } we hook the TransactionEventHandler to the Deuce api via System properties, the same way we hook transaction manager (lsa, tl2, ...) I made it more generic because i think that, with my persistence implementation, it would not be possible to persist correctly all type of transaction and that persistence would not be achieved by Deuce itself. So this api modification can serve more purposes.
    we define a Method class that hold all informations we need on origin method. I added getAnnotations() method because i want to have the flexibility to allow other annotations that the handler can use (like @Persist) class Method { /* return method name */ public String getName(); /* return method parameters values or an empty array if none */ public Object[] getParameters(); /* return run time annotations or an empty array */ public Annotation[] getAnnotations(); } we define a TransactionEvent class class TransactionEvent { public static enum Type { INIT, COMMIT, ROLLBACK, RETRY, /* HOOK to allow, just before commiting, an external agent to do something that if throws an exception will rollback the transaction. */ HOOK }; public Type getType(); public Method getOriginMethod(); /* return the origin object (a class if method call was static) */ public Object getOriginObject(); } we define a TransactionEventHandler interface interface TransactionEventHandler { /* called after event in Deuce */ void handle(TransactionEvent event); } we hook the TransactionEventHandler to the Deuce api via System properties, the same way we hook transaction manager (lsa, tl2, ...) I made it more generic because i think that, with my persistence implementation, it would not be possible to persist correctly all type of transaction and that persistence would not be achieved by Deuce itself. So this api modification can serve more purposes.
  • Nov 12, 2009
    r224 (1. Uniquify the rest of the databases. 2. Do the advice thin...) committed by arie.zilberstein   -   1. Uniquify the rest of the databases. 2. Do the advice thing only in field accesses and array accesses. No need for all the other opcodes. 3. The advisor is actually for TL2 so move it there. For example: a newly-allocated object might require logging in non-TL2 STMs. Should verify if there are any real examples.
    1. Uniquify the rest of the databases. 2. Do the advice thing only in field accesses and array accesses. No need for all the other opcodes. 3. The advisor is actually for TL2 so move it there. For example: a newly-allocated object might require logging in non-TL2 STMs. Should verify if there are any real examples.
  • Nov 12, 2009
    r223 (1. Uniquify code points. Previously the method signature was...) committed by arie.zilberstein   -   1. Uniquify code points. Previously the method signature was ignored, so overloaded methods were considered like they were the same one. 2. Fix Pruner3 - use LeakyConstructorDatabase to detect leaky ctors. The previous implementation didn't treat calls to this() and super() correctly and missed leaks. 3. Let the immutability optimizer implement the same interface like the others.
    1. Uniquify code points. Previously the method signature was ignored, so overloaded methods were considered like they were the same one. 2. Fix Pruner3 - use LeakyConstructorDatabase to detect leaky ctors. The previous implementation didn't treat calls to this() and super() correctly and missed leaks. 3. Let the immutability optimizer implement the same interface like the others.
  • Nov 11, 2009
    r222 (Fix bug in array transformation + update from trunk) committed by arie.zilberstein   -   Fix bug in array transformation + update from trunk
    Fix bug in array transformation + update from trunk
  • Nov 11, 2009
    r221 (Fix compilation problems) committed by arie.zilberstein   -   Fix compilation problems
    Fix compilation problems
  • Nov 09, 2009
    r220 (Add TransactionManager and Advisor) committed by gkorland   -   Add TransactionManager and Advisor
    Add TransactionManager and Advisor
  • Nov 07, 2009
    r219 (A method is eligible for LastFieldActivityAnalysis only if i...) committed by arie.zilberstein   -   A method is eligible for LastFieldActivityAnalysis only if it is atomic starter. Because other methods may be contained in transactions.
    A method is eligible for LastFieldActivityAnalysis only if it is atomic starter. Because other methods may be contained in transactions.
  • Oct 31, 2009
    r218 (Issue 2 Fix external field holder to include a super class.) committed by gkorland   -   Issue 2 Fix external field holder to include a super class.
    Issue 2 Fix external field holder to include a super class.
  • Oct 31, 2009
    r217 (Add tl2cm scripts) committed by yoavon.cohen   -   Add tl2cm scripts
    Add tl2cm scripts
  • Oct 31, 2009
    r216 (Add tl2cm scripts) committed by yoavon.cohen   -   Add tl2cm scripts
    Add tl2cm scripts
  • Oct 31, 2009
    r215 (Add tl2cm scripts) committed by yoavon.cohen   -   Add tl2cm scripts
    Add tl2cm scripts
  • Oct 31, 2009
    r214 (Add tl2cm scripts) committed by yoavon.cohen   -   Add tl2cm scripts
    Add tl2cm scripts
  • Oct 31, 2009
    r213 (Add tl2cm scripts) committed by yoavon.cohen   -   Add tl2cm scripts
    Add tl2cm scripts
  • Oct 29, 2009
    r212 (Let method invocations destroy the optimization opportunity ...) committed by arie.zilberstein   -   Let method invocations destroy the optimization opportunity of LastFieldActivity
    Let method invocations destroy the optimization opportunity of LastFieldActivity
  • Oct 27, 2009
    r211 ([No log message]) committed by gkorland   -   [No log message]
    [No log message]
  • Oct 27, 2009
    r210 (Issue 41 Adding the TL2CM first version) committed by gkorland   -   Issue 41 Adding the TL2CM first version
    Issue 41 Adding the TL2CM first version
  • Oct 27, 2009
    issue 41 (Add TL2CM) Status changed by gkorland   -  
    Status: Fixed
    Status: Fixed
  • Oct 27, 2009
    issue 41 (Add TL2CM) reported by gkorland   -   Add a version of TL2 with a contention manager.
    Add a version of TL2 with a contention manager.
  • Oct 24, 2009
    r209 (Integrate soot with optimizations and analyses 5) committed by arie.zilberstein   -   Integrate soot with optimizations and analyses 5
    Integrate soot with optimizations and analyses 5
  • Oct 24, 2009
    r208 (Integrate soot with optimizations and analyses 4) committed by arie.zilberstein   -   Integrate soot with optimizations and analyses 4
    Integrate soot with optimizations and analyses 4
  • Oct 24, 2009
    r207 (Integrate soot with optimizations and analyses 3) committed by arie.zilberstein   -   Integrate soot with optimizations and analyses 3
    Integrate soot with optimizations and analyses 3
  • Oct 24, 2009
    r206 (Integrate soot with optimizations and analyses 2) committed by arie.zilberstein   -   Integrate soot with optimizations and analyses 2
    Integrate soot with optimizations and analyses 2
  • Oct 24, 2009
    r205 (Integrate soot with optimizations and analyses 1) committed by arie.zilberstein   -   Integrate soot with optimizations and analyses 1
    Integrate soot with optimizations and analyses 1
  • Oct 24, 2009
    r204 (Add soot jars) committed by arie.zilberstein   -   Add soot jars
    Add soot jars
  • Oct 24, 2009
    r203 (Introduce deuce optimizations based on soot and custom stati...) committed by arie.zilberstein   -   Introduce deuce optimizations based on soot and custom static analyses
    Introduce deuce optimizations based on soot and custom static analyses
  • Oct 23, 2009
    r202 (Catch up with trunk) committed by arie.zilberstein   -   Catch up with trunk
    Catch up with trunk
  • Oct 22, 2009
    r201 (Clean code) committed by gkorland   -   Clean code
    Clean code
  • Oct 22, 2009
    issue 40 (Faild on switch block when Enum is used.) Status changed by gkorland   -  
    Status: Verified
    Status: Verified
 
Hosted by Google Code