Skip to content

Release 10/2026

Latest

Choose a tag to compare

@fabianbs96 fabianbs96 released this 24 Oct 10:47
Immutable release. Only release title and notes can be modified.
35aa48e

Dear community,

today we are releasing PhASAR v2610. These are the main changes:

  • New high-performance bottom-up IFDS solver (MonoIFDSSolver), presented at ECOOP as "Scaling Bottom-Up IFDS Taint Analysis with Optimized Data-Flow Encoding" https://doi.org/10.4230/LIPIcs.ECOOP.2026.23
  • New Andersen-style pointer analysis with on-the-fly call-graph construction and selective context-sensitivity (AndersenOTFAA)
  • New high-performance solver for IFDS problems, DFISolver, based on DFS-interval reachability (experimental)
  • New solver for IFDS/IDE problems, WPDSSolver, based on weighted pushdown systems (experimental)
  • Add support for SparseIFDS/SparseIDE in IterativeIDESolver
  • Add support for all LLVM versions from 16 to 22.1
  • Add support for path-tracking in IterativeIDESolver
  • Model extern callbacks in ICFG to improve soundness (contributed by @e345ee)
  • New example showing how to efficiently export a call-graph (contributed by @WizardBornov)
  • Global-initializers handling now supports library code
  • compose and join no longer have to be declared in edge-functions if the problem provides extend and combine
  • Devirtualize IFDS/IDE problem (see BreakingChanges.md)
  • Add LIT Testing:
    • Tests can now express expectations inline directly in terms of tool-output
    • Bug reports can now be directly converted to test cases
  • Removed deprecated APIs (see BreakingChanges.md)
  • Several bugfixes and smaller changes

For details please have a look at the merging PR #870.

For the next release (probably in March/April 2027) we plan to update the default configuration for both HelperAnalyses and phasar-cli as breaking changes;
this will allow us to deprecate old analyses and improve the experience for most users.

Thank you for using PhASAR. We appreciate your feedback, bug reports and contributions.
The PhASAR team