Segmentation fault when reading erroneous mechanisms without transport data

[thomasfiala]

When reading an erroneous mechanism (both .cti or .xml) that doesn't include transport parameters but shows a transport = 'Mix' or transport = 'Multi'` flag, Cantera returns a segmentation fault. Probably, there should be a warning instead.

Example: Mechanism:

units(length = "cm", time = "s", quantity = "mol", act_energy = "cal/mol")

ideal_gas(
      name = "gas",
      elements = " H ",
      species = """ H """,
      reactions = "all",
      transport="Mix",  # also returns segfault for "Multi"
      initial_state = state(temperature = 300.0, pressure = OneAtm)
      )

#-------------------------------------------------------------------------------
#  Species data 
#-------------------------------------------------------------------------------

species(
    name = "H",
    atoms = " H:1 ",
    thermo = (
       NASA( [  300.00,  1000.00], [  2.500000000E+00,   0.000000000E+00, 
                0.000000000E+00,   0.000000000E+00,   0.000000000E+00,
                2.547163000E+04,  -4.601176000E-01] ),
       NASA( [ 1000.00,  5000.00], [  2.500000000E+00,   0.000000000E+00, 
                0.000000000E+00,   0.000000000E+00,   0.000000000E+00,
                2.547163000E+04,  -4.601176000E-01] )
             ),
    #transport=gas_transport(geom='atom', diam=2.05, well_depth=145.0)
    #no error is returned if transport data is not commented out
    )

[bryanwweber]

Hi Thomas,
Can you provide a few more details to reproduce the segfault? I use mechanisms all the time that have no transport info, and I have no trouble. I use them for homogeneous cases, so no Transport is necessary - is the segfault showing when you use this kind of mechanism in a 1D case?

[thomasfiala]

The problem is that the mechanism shows a transport='Mix' or transport='Multi' flag. If you comment that out, the mechanism loads just fine. (I updated the description for this information.)

The segfault occurs when loading the mechanism. In python, a simple gas = ct.Solution('mech_test.cti') will do.

The error was brought up by a colleague of me who used an erroneous chemkin mechanism for a 1D flame.