We can characterise the state of a concurrent computation by considering the ordering of dependent events. This is a partial order: independent events can be performed in any order without affecting the result. DPOR is a technique for computing these partial orders at run-time, and only testing one total order for each partial order. This cuts down the amount of work to be done significantly. In particular, this package implemented bounded partial-order reduction, which is a further optimisation. Only schedules within some bound are considered.
executions /within the bound/!
Caution: The fundamental assumption behind DPOR is that the only source of nondeterminism in your program is the scheduler. Or, to put it another way, if you execute the same program with the same schedule twice, you get the same result. If you are using this library in combination with something which performs I/O, be very certain that this is the case!
See the <https://github.com/barrucadu/dejafu README> for more details.
For details on the algorithm, albeit presented in a very imperative way, see /Bounded partial-order reduction/, K. Coons, M. Musuvathi, and K. McKinley (2013), available at <http://research.microsoft.com/pubs/202164/bpor-oopsla-2013.pdf>.
|Package Version||Update ID||Released||Package Hub Version||Platforms||Subpackages|
|0.2.0.0-bp150.2.4 info||GA Release||2018-08-01||15||
|0.2.0.0-bp150.2.6 info||GA Release||2018-07-31||15||
|0.2.0.0-bp150.2.7 info||GA Release||2018-07-30||15||