07.01.2013 Views

Lecture Notes in Computer Science 3472

Lecture Notes in Computer Science 3472

Lecture Notes in Computer Science 3472

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

5 Preorder Relations 143<br />

for free. Indeed, the tests <strong>in</strong> the sets O1 and O2 described <strong>in</strong> the previous sections<br />

are sequential, <strong>in</strong> the sense that unions always occur between a test whose result<br />

that is immediately available (Succ or Fail) and some other, possibly longer<br />

test. In test<strong>in</strong>g preorders as well as <strong>in</strong> failure trace preorder we need to copy the<br />

process while it runs; <strong>in</strong>deed, we may need to comb<strong>in</strong>e the outcomes of two (or<br />

more) different runs of the process, which means that we need to run two copies<br />

of the process to obta<strong>in</strong> these outcomes <strong>in</strong>dependently from each other. Because<br />

of the sequential tests used by refusal preorder copy<strong>in</strong>g is no longer necessary<br />

(but it becomes necessary once more <strong>in</strong> failure trace preorder). This be<strong>in</strong>g said,<br />

the def<strong>in</strong>ition of the must operator from refusal preorder implies that processes<br />

need to be copied anyway (s<strong>in</strong>ce we have to apply many tests on them), so the<br />

failure trace test<strong>in</strong>g scenario is not that bad after all.<br />

5.8 Fair Test<strong>in</strong>g<br />

Recall the processes depicted <strong>in</strong> Figure 5.6 on page 133 and our back and forth<br />

argument that they should be considered equivalent (or not). When we considered<br />

them under the test<strong>in</strong>g preorder, s and t were not equivalent, whereas they<br />

are so under the other preorders. Test<strong>in</strong>g preorder, with its habit that the presence<br />

of divergence may ru<strong>in</strong> a test, will differentiate between these two processes<br />

as opposed to all the other preorders we have seen so far. As we mentioned,<br />

whether such a behavior is a good or bad th<strong>in</strong>g depends on one’s op<strong>in</strong>ion about<br />

divergences.<br />

For those who prefer to ignore divergences as long as there is a hope that<br />

the process will cont<strong>in</strong>ue with its visible operation, i.e., for those who prefer to<br />

consider the processes shown <strong>in</strong> Figure 5.6 equivalent, fair test<strong>in</strong>g is available<br />

[BRV95].<br />

We have the same test<strong>in</strong>g scenario for fair test<strong>in</strong>g as the one used <strong>in</strong> Section<br />

5.5, except that the operator must is enhanced, such that it chooses a visible<br />

action whenever such an action is available. With the same set O of observers as<br />

the one used to def<strong>in</strong>e the test<strong>in</strong>g preorder, the new operator fmust is def<strong>in</strong>ed<br />

as follows:<br />

p fmust o iff for any σ ∈ Act ∗ and o ′ ∈Owith o = σo ′ , it holds that:<br />

obs(o ′ , p ′ ) = obs(o, p) forsomep ′ ∈ Q, p σ ⇒ p ′ , implies that there<br />

exists a ∈ Act ∪{Succ} such that o ′ = ao ′′ , o ′′ ∈O∪{ε}.<br />

The preorder ⊑fmust, as well as the equivalence �fmust <strong>in</strong>duced by the operator<br />

fmust are def<strong>in</strong>ed <strong>in</strong> the usual manner.<br />

The operator fmust is the “fair” variant of the operator must of test<strong>in</strong>g<br />

preorder l<strong>in</strong>eage. It ignores the divergences as long as there is a visible action (a<br />

<strong>in</strong> the above def<strong>in</strong>ition) accessible to the observer. The follow<strong>in</strong>g characterization<br />

of ⊑fmust <strong>in</strong> terms of other preorders is easily obta<strong>in</strong>ed from the results presented<br />

elsewhere [BRV95]:

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!