More often than not, testing software consumes a large portion of the development budget, however we frequently see cases where unit and integration tests fail to uncover critical errors that appear once the software is deployed. Most testing techniques revolve around specifying a collection of execution sequences that check the expected against the actual behavior. A problem with this is that the number of possible execution sequences is huge, and therefore only a very small portion of these would be covered by test cases that are specified as a sequence of steps. The second problem is that, with the goal of increasing coverage and prevent regression bugs a large number of test cases is written, which eats up the development budget.
Model-based testing is a technique for writing tests, where a model of the system behavior is made a a high-level of abstraction, and then the system-under test is tested against this the expected behavior as specified by the model. Model-based testing relies on different algorithms for generating test cases from models, which allows to achieve a much higher test coverage than standard testing techniques, while requiring only a fraction of the code.
TorXakis is such a model-based testing tool, that has been used to verify large scale systems in well-know high tech companies. This tool is entirely written in Haskell, and its code is available on Github under a BSD3 license.
Since July last year, a lot of effort was put into taking TorXakis from a prototype to an industrial grade tool. Some of the improvements made include:
setup of continuous integration (Windows/Linux), including hlint quality metrics via code climate.
release of macOS and Windows installers.
addition of integration tests and benchmarks.
improvements in performance.
architectural simplifications.
In addition, there is ongoing work in a new command line interface and a new compiler for the TorXakis language.
A year is almost gone, and there are a lot of interesting challenges ahead to make TorXakis a tool that can be used in production, so we welcome the contributions of anybody interested in the topic.
Hey everyone! We're working on TorXakis.
More often than not, testing software consumes a large portion of the development budget, however we frequently see cases where unit and integration tests fail to uncover critical errors that appear once the software is deployed. Most testing techniques revolve around specifying a collection of execution sequences that check the expected against the actual behavior. A problem with this is that the number of possible execution sequences is huge, and therefore only a very small portion of these would be covered by test cases that are specified as a sequence of steps. The second problem is that, with the goal of increasing coverage and prevent regression bugs a large number of test cases is written, which eats up the development budget.
Model-based testing is a technique for writing tests, where a model of the system behavior is made a a high-level of abstraction, and then the system-under test is tested against this the expected behavior as specified by the model. Model-based testing relies on different algorithms for generating test cases from models, which allows to achieve a much higher test coverage than standard testing techniques, while requiring only a fraction of the code.
TorXakis is such a model-based testing tool, that has been used to verify large scale systems in well-know high tech companies. This tool is entirely written in Haskell, and its code is available on Github under a BSD3 license.
Since July last year, a lot of effort was put into taking TorXakis from a prototype to an industrial grade tool. Some of the improvements made include:
In addition, there is ongoing work in a new command line interface and a new compiler for the TorXakis language.
A year is almost gone, and there are a lot of interesting challenges ahead to make TorXakis a tool that can be used in production, so we welcome the contributions of anybody interested in the topic.
Further reading: github.com/TorXakis/TorXakis/
Cheers!