XSB problems

[Bart Bogaerts]

Na mergen van laatste pull request https://bitbucket.org/krr/idp/pull-request/301/xsb-fixes

gaat er 't een en 't ander mis in het testframework gerelateerd aan xsb

Op zijn minst twee falende tests:

Failed: CalculateDefinitionsWithXSB/CalculateDefinitionsWithXSBTest.CalculatesDefinitionWithXSB/52
Testing /home/bartb/software/idp/tests/definitionWithXSBtests/twovalued_opens/knapsack.idp
FAIL! Calculating the definition with XSB resulted in no model, while without XSB it did have a model.
/home/bartb/software/idp/tests/mxtests/definitionWithXSBtests.cpp:80: Failure
Value of: Status::SUCCESS
  Actual: success
Expected: result
Which is: failed

en

[==========] RefinesDefinitionWithXSB/RefineDefinitionsWithXSBTest.RefinesDefinitionWithXSB/63
Testing /home/bartb/software/idp/tests/definitionWithXSBtests/twovalued_opens/gcd.idp
Refined definitions with XSB...Out of resources
Shutdown failed, aborting.
Aborted (core dumped)

Mogelijks gaat er dus nog meer mis...

aub: run make check voor je een pull request mergt!

[Bart Bogaerts]

Revert "Merged in xsbupdate (pull request #301) "

This reverts commit 5e93b4725c5aa588fdc181d4cf0814809f26fb36, reversing changes made to 820a1ef3031b0d7baee28bd6b8c9c39e3f217174.

See #838

→ <<cset dbdcc95a941f>>

[Bart Bogaerts]

Changes uit die pull request zijn terug uit master en zitten in https://bitbucket.org/krr/idp/commits/branch/xsb_fixes_dirty

[Joachim Jansen]

Ik kan niet meteen vinden waar het probleem zit. Ik krijg segfaults omdat gcd.idp in een timeout gaat en het "killen" zorgt voor de segfault.

Ik kijk er volgende week naar.

[Joachim Jansen]

The first failing test case (knapsack.idp) was a result of a faulty commit in this branch. The second failing test case (gcd.idp) is a result of a timeout. This timeout is caught during XSB's run and is not handled properly. This is then reduced to issue #672

Something to think about: I looked at why gcd.idp is so slow and I found that a relative simple definition is transformed into a gigantic one. This might be worth looking into. The example is given below:

{
    dist(a,b)= xDist(a,b)*xDist(a,b)+yDist(a,b)*yDist(a,b).
}

Is transformed into

{
    ! a[El] b[El] var3339[El] : dist(a[El],b[El],var3339[El]) <- 
        (
            (? var3340[s1..3] var3341[s1..3] var3342[s1..9] var3343[s1..3] var3344[s1..3] var3345[s1..9] : 
                (xDist(a[El],b[El],var3340[s1..3]) & 
                (xDist(a[El],b[El],var3341[s1..3]) & 
                (
                    *(var3340[s1..3],var3341[s1..3],var3342[s1..9]) & 
                    *(
                        yDist(a[El],b[El],var3343[s1..3]) & 
                        *(
                            yDist(a[El],b[El],var3344[s1..3]) & 
                            *(var3343[s1..3],var3344[s1..3],var3345[s1..9]) &
                            +(var3342[s1..9],var3345[s1..9],var3339[El]))
                    ) & 
            (? var3346[s1..3] var3347[s1..3] var3348[s1..9] var3349[s1..3] var3350[s1..3] var3351[s1..9] var3352[s2..18] : 
                (
                    xDist(a[El],b[El],var3346[s1..3]) & 
                    xDist(a[El],b[El],var3347[s1..3]) & 
                    *(var3346[s1..3],var3347[s1..3],var3348[s1..9]) & 
                    yDist(a[El],b[El],var3349[s1..3]) & 
                    yDist(a[El],b[El],var3350[s1..3]) & 
                    *(var3349[s1..3],var3350[s1..3],var3351[s1..9]) & 
                    +(var3348[s1..9],var3351[s1..9],var3352[s2..18]) & 
                    El(var3352[s2..18])
                )
            )
        ).
}