Skip to content

Mungo comparison

João Mota edited this page Jul 29, 2021 · 3 revisions

This table compares the checking capabilities between Mungo and our tool. Each line is justified by some tests, which include a README.md file with the results each tool outputs. On each test, the correct output is characterized by no errors being reported in Ok.java files and errors being reported in NotOk.java files.

Features Mungo Toolset Java Typestate Checker Tests
Basic checking ✔️ ✔️ basic-checking
Decisions on enumeration values ✔️ ✔️ basic-checking
Decisions on boolean values ✔️ boolean-decision
@State ✔️ state-refinement
Nullness checking 🟨 [1] ✔️ nullness-checking
Linearity checking 🟨 [2, 3] ✔️ linearity-checking, linearity-checking-corner-cases
Force protocol completion 🟨 [4] ✔️ protocol-completion, protocol-completion-corner-cases
Class analysis ✔️ class-analysis
Protocol definitions for libraries ✔️ iterator-attempt1
Droppable objects ✔️ droppable-objects
Generics generics
Collections auction
Improved flow analysis
Recognition of decisions with equality checks
✔️ [5] iterator-attempt[2, 3, 4, 5, 6, 7]
original-mungo-flow-issue-1
original-mungo-flow-issue-2
  1. Some errors are reported but they are a bit cryptic and obj != null refinement does not exist.
  2. Some errors are reported but they are a bit cryptic.
  3. Some corner cases are not handled, namely, if an object is moved to a lambda.
  4. Some corner cases are not handled, namely, if an object without protocol holds a reference to an object with protocol, the protocol completion of the second one if not ensured.
  5. We recognize, for example, that while(true) is an infinite loop. This is useful for recognition of code that might have been automatically generated by a toolchain, like StMungo.
Clone this wiki locally