- remove extraneous printing
- fixes for parse errors in JS and HTML
- fixes for handling parse errors in JS and HTML
- update comments
- Change BitVectorRepository to use LinkedLists
- improve javadoc
- fix for for in contexts for NEVER case
- missing VectorKill println method
- Annotation support
- Properly fix path-with-spaces bug.
- fix bug involving paths with spaces
- add a simple driver for building a call graph via a scope file
- Properly return null as default constructor of an array.
- organize imports
- better handling of missing bytecodes
- javadoc
- test fix
- small Javadoc fix
- added date-property.js
- 1) added InstanceKey.getCreation sites and its implementations 2) fixes for issues with keys representing dynamic properties i) all properties are converted to strings,
- publicize method makeClasspath(). deprecate quoteStringIfNeeded()
- organize imports
- javadoc
- renamed classes to make relationship to mod-ref analysis clearer
- add support for lexical writes
- Code to compute transitive lexical accesses of methods.
- extract some generally useful code from ModRef
- Generate proper InstanceFieldKeys for property accesses with Numbers.
- rewrite to make hardest test appear last
- fix test to properly check reachability
- add an array test that doesn't quite work
- add method to get a PointerKey for a global
- compare FieldValueDispatch objects based on CGNode as well
- Handle duplicate field names between subclass and superclass.
Specifically, rewrite ProgressMaster to not depend on Eclipse, and move
to com.ibm.wala.util. Now, we can use timeout-based code in packages
without introducing an Eclipse dependency.
- basic compatibility with Java 7 (i.e., don't crash immediately)
- Added utility class for converting call graphs to JSON.
- add edgeExists CLI option to check if some edge exists in the call graph
Due to constant parameters, we can't assume that constraints don't need
to be generated when we've already seen a target at a call site (since
the previous constraints may have only passed certain constant parameter
values, rather than all parameter values). Add a check to handle these
cases correctly.
The issue was that it's possible for multiple invoke instructions with
different actual parameters to be associated with a single CallSiteReference.
In this case, the invariant parameters for each invoke instruction may differ.