Cactus does not detect circular schedule items

Issue #910 closed
Roland Haas created an issue

Right now none of the three schedule items below (which should all be flagged as faulty I think), triggers any warning during compile or run time:

schedule group ordertest at initial
{
} "test schedule sorter"

schedule ADMBase_CartesianMinkowski in ordertest after ADMBase_LapseOne
{
  LANG: C
} "Set the metric and extrinsic curvature to cartesian minkowski values"

schedule ADMBase_LapseOne in ordertest after ADMBase_CartesianMinkowski
{
  LANG: C
} "Set the lapse to 1 at all points"

SCHEDULE ADMBase_SetShiftStateOn in ordertest after ADMBase_SetShiftStateOn
{
  LANG: C
} "Set the shift_state variable to 1"

I attach: * a patch against ADMBase/schedule.ccl to demonstrate the output * stdout when one runs Exact/Schwarzschild_EF with the faulty schedule.ccl * a patch against src/schedule/ScheduleSorter.c and Creator that adds error checks for this situation (the part inside of TEST_SCHEDULECREATOR is for the stand-alone test inside to compile)

Keyword:

Comments (6)

  1. Erik Schnetter
    • changed status to open
    • removed comment

    The first patch is okay to apply.

    The second patch does not say to which thorn it should be applied. The patch is okay if it is applied to thorn in CactusTest (but not to the flesh or e.g. an Einstein Toolkit thorn).

  2. Roland Haas reporter
    • removed comment

    ordertest_schedule.ccl.patch is not intended to be applied. It was only provided so that others could test the reported problem. It is a patch against ADMBase.

    I applied the first patch detect_circular_dependencies.patch as rev 4832 of the flesh.

  3. Log in to comment