Section: New Software and Platforms

Others

Below we list some software that has been developed, or is under development, in Focus.

• Deadlock analysis (http://df4abs.nws.cs.unibo.it/ ).

  We have prototyped a framework for statically detecting deadlocks in a concurrent object-oriented language with asynchronous method calls and cooperative scheduling of method activations (the language is inspired by the ABS language developed in the EU project HATS and currently extended with primitives for cloud-computing in the EU project ENVISAGE). Since this language features recursion and dynamic resource creation, deadlock detection is extremely complex and state-of-the-art solutions either give imprecise answers or do not scale. In order to augment precision and scalability we propose a modular framework that allows several techniques to be combined. The basic component of the framework is a front-end inference algorithm that extracts abstract behavioural descriptions of methods, called contracts, which retain resource dependency information. Then these contracts are analysed by a back-end that uses a fix-point technique to derive in a deterministic way the deadlock information.

• CaReDeb (http://www.cs.unibo.it/caredeb ).

  Reversible debugging provides developers with a way to execute their applications both forward and backward, seeking the cause of an unexpected or undesired event. We have developed CaReDeb, the first prototype of a causal-consistent reversible debugger. Causal consistent here means that independent actions are undone independently, while dependent actions are undone in reverse order. This allows the programmer to concentrate on the threads responsible of the bug, independently of the actual interleaving. CaReDeb provides primitives that given a misbehaviour, e.g., a variable has not the expected value, allow one to go back to the action responsible for it, e.g., the one that assigned the wrong value to the variable. Notably, the programmer has no need to know which thread the action belongs to, since this is found automatically by the debugger. The procedure can be iterated till the bug is found. CaReDeb targets a fragment of the language Oz, which is at the basis of Mozart. The considered fragment provides functional variables, procedures, threads, and asynchronous communication via ports.

• AIOCJ (http://www.cs.unibo.it/projects/jolie/aiocj.html ).

  AIOCJ is a framework for programming adaptive distributed systems based on message passing. AIOCJ comes as a plugin for Eclipse, AIOCJ-ecl, allowing to edit descriptions of distributed systems as adaptive interaction-oriented choreographies (AIOC). From interaction-oriented choreographies the description of single participants can be automatically derived. Adaptation is specified by rules allowing to replace predetermined parts of the AIOC with a new behaviour. A suitable protocol ensures that all the participants are updated in a coordinated way. As a result, the distributed system follows the specification given by the AIOC under all changing sets of adaptation rules and environment conditions. In particular, the system is always deadlock-free. AIOCJ can interact with external services, seen as functions, by specifying their URL and the protocol they support (HTTP, SOAP, ...). Deadlock-freedom guarantees of the application are preserved provided that those services do not block.

• METIS (https://github.com/aeolus-project/metis )

  As partners of the Aeolus project we have developed a tool for the automatic synthesis of deployment plans. A deployment plan is a sequence of actions that, when performed, allows the deployment of a given configuration of components. METIS (Modern Engineered Tool for Installing Software systems) is a tool that enables one to automatically generate a deployment plan, starting from a description of the configuration following the Aeolus model. The software is open source. It is written entirely in OCaml and is about 3.5K lines of source code. The tool is based on theoretical results that guarantee its soundness and completeness, while maintaining polynomial computational complexity. METIS already showed its effectiveness in practice by handling synthesized problem instances with hundreds of components in less than a minute. We are currently validating Metis in a production environment by integrating it in Armonic, an infrastructure for cloud application deployment in OpenStack cloud systems developed by the Mandriva company.

• SUNNY-CP (https://github.com/jacopoMauro/sunny-cp )

  Within the Constraint Programming (CP) paradigm, a portfolio approach enables to combine a number of different constraint solvers in order to create a globally better solver, dubbed a portfolio solver. After several empirical evaluations (e.g., [22] , [23] , [24] ) we have decided to develop SUNNY-CP, a portfolio solver for solving both Constraint Satisfaction Problems and Constraint Optimization Problems. The goal of SUNNY-CP is to provide a flexible, configurable, and usable CP portfolio solver that can be set up and executed just like a regular individual CP solver. To the best of our knowledge, SUNNY-CP is the only sequential portfolio solver able to solve generic CP problems, and it was the only portfolio solver that attended the MiniZinc Challenge 2014 (i.e., the only active international competition to evaluate the performance of CP solvers). (SUNNY-CP performed very well, ranking 4th in the competition and receiving an honourable mention by the challenge organizers.) The application of SUNNY-CP in the optimization problems defined within the Aeolus project have lead to time improvements beyond an order of magnitude. SUNNY-CP is mainly written in Python, and we are currently enhancing the tool in order to make it more usable, flexible, and parallel (i.e., able to properly exploit multiple cores).

The sofware below have not undergone substantial modifications during 2014.

• Croll-pi Interpreter (http://proton.inrialpes.fr/~mlienhar/croll-pi/implem/ ). Croll-pi is a concurrent reversible language featuring a rollback operator to undo a past action (together with all the actions depending on it), and a compensation mechanism to avoid cycling by redoing the same action again and again. We have developed an interpreter for croll-pi using Maude.

• IntML is a functional programming language guaranteeing sublinear space bounds for all programs [51] . See the Activity Reports of previous years (in particular 2010) for more details.

• Lideal (http://lideal.cs.unibo.it/ ) is an experimental tool implementing type inference for dependently linear type systems. The tool reduces the problem of evaluating the complexity of PCF (i.e. functional programs with primitive integers and recursive definitions) to checking a set of first-order inequalities for validity. The latter can then be handled through SMT solvers or put in a form suitable for managing them with tools such as CoQ. See the Activity Reports of previous years (in particular 2010) for more details.