Regresion on the 555 example

[guitorri]

These examples work fine with 0.0.18.

• 555_examples_prj

The two top schematics cause segfaults with current master.

Traceback:

NOTIFY: TR1: solving transient netlist
Process 89081 stopped
* thread #1: tid = 0x56f0d9, 0x000000010016df53 libqucs.0.dylib`qucs::history::seek(this=0x0000000100e09cb8, tval=-0.0000000050000000000000001, l=0, r=-1, diff=0x00007fff5fbfd758, idx=-1) + 131 at history.cpp:129, queue = 'com.apple.main-thread', stop reason = EXC_BAD_ACCESS (code=1, address=0x0)
    frame #0: 0x000000010016df53 libqucs.0.dylib`qucs::history::seek(this=0x0000000100e09cb8, tval=-0.0000000050000000000000001, l=0, r=-1, diff=0x00007fff5fbfd758, idx=-1) + 131 at history.cpp:129
   126    int i = (l + r) / 2;
   127    if (l == r)
   128      return i;
-> 129    nr_double_t v = (*this->t)[i];
   130    nr_double_t d = v - tval;
   131    if (fabs (d) < diff) {
   132      // better approximation
(lldb) bt
* thread #1: tid = 0x56f0d9, 0x000000010016df53 libqucs.0.dylib`qucs::history::seek(this=0x0000000100e09cb8, tval=-0.0000000050000000000000001, l=0, r=-1, diff=0x00007fff5fbfd758, idx=-1) + 131 at history.cpp:129, queue = 'com.apple.main-thread', stop reason = EXC_BAD_ACCESS (code=1, address=0x0)
  * frame #0: 0x000000010016df53 libqucs.0.dylib`qucs::history::seek(this=0x0000000100e09cb8, tval=-0.0000000050000000000000001, l=0, r=-1, diff=0x00007fff5fbfd758, idx=-1) + 131 at history.cpp:129
    frame #1: 0x000000010016de4f libqucs.0.dylib`qucs::history::nearest(this=0x0000000100e09cb8, tval=-0.0000000050000000000000001, interpolate=true) + 239 at history.cpp:113
    frame #2: 0x0000000100079d6e libqucs.0.dylib`qucs::circuit::getV(this=0x0000000100e19250, port=1, t=-0.0000000050000000000000001) + 62 at circuit.cpp:940
    frame #3: 0x00000001002962c7 libqucs.0.dylib`digital::getVin(this=0x0000000100e19250, input=0) + 71 at digital.cpp:76
    frame #4: 0x00000001002964eb libqucs.0.dylib`digital::calcTransferX(this=0x0000000100e19250, input=0) + 459 at digital.cpp:86
    frame #5: 0x000000010029658b libqucs.0.dylib`digital::calcTransfer(this=0x0000000100e19250, input=0) + 27 at digital.cpp:91
    frame #6: 0x0000000100297b9e libqucs.0.dylib`logicnor::calcOutput(this=0x0000000100e19250) + 318 at nor.cpp:45
    frame #7: 0x0000000100297635 libqucs.0.dylib`digital::calcTR(this=0x0000000100e19250, t=0) + 325 at digital.cpp:180
    frame #8: 0x0000000100150093 libqucs.0.dylib`qucs::trsolver::calcTR(self=0x0000000100e16900) + 83 at trsolver.cpp:740
    frame #9: 0x00000001000b47e3 libqucs.0.dylib`qucs::nasolver<double>::calculate(this=0x0000000100e16900) + 51 at nasolver.h:72
    frame #10: 0x00000001000b2c4f libqucs.0.dylib`qucs::nasolver<double>::solve_once(this=0x0000000100e16900) + 47 at nasolver.cpp:143
    frame #11: 0x00000001000ae45f libqucs.0.dylib`qucs::nasolver<double>::solve_nonlinear(this=0x0000000100e16900) + 1263 at nasolver.cpp:480
    frame #12: 0x000000010015050c libqucs.0.dylib`qucs::trsolver::corrector(this=0x0000000100e16900) + 28 at trsolver.cpp:574
    frame #13: 0x000000010014ec1a libqucs.0.dylib`qucs::trsolver::solve(this=0x0000000100e16900) + 2138 at trsolver.cpp:281
    frame #14: 0x000000010009235c libqucs.0.dylib`qucs::net::runAnalysis(this=0x0000000100e08190, err=0x00007fff5fbfee24) + 1628 at net.cpp:264
    frame #15: 0x0000000100003a2d qucsator`main(argc=5, argv=0x00007fff5fbff9b8) + 8957 at ucs.cpp:251
    frame #16: 0x00007fff81edf5fd libdyld.dylib`start + 1
    frame #17: 0x00007fff81edf5fd libdyld.dylib`start + 1
(lldb) 

[guitorri]

@bastien-roucaries would you take a look at this?

[bastien-roucaries]

Could we upack example on git tree ? I will allow to do regression testing on it.

Bastien

On Fri, Jan 30, 2015 at 4:29 PM, Guilherme Brondani Torri < notifications@github.com> wrote:

@bastien-roucaries https://github.com/bastien-roucaries would take a look at this?

— Reply to this email directly or view it on GitHub https://github.com/Qucs/qucs/issues/201#issuecomment-72218145.

[in3otd]

...since it seems we will soon start some work on the transient solver, it will make sense to add several known currently-failing circuits to the test suite. Maybe leave them disabled by default and add a switch to include also (or only) them for testing during the development. I have already collected a few from the various issues and discussions, will later post a message to qucs-devel to ask for some more examples which fail the transient simulations

[guitorri]

There are some numerical differences with respect to 0.0.17 and 0.0.18. I am trying to bisect it.

[guitorri]

Sorry for the noise. The issue is not with the 555 example...