#636: PITTNullCode does not compile on IBM P7 using the ibm compilers. -----------------------------------+---------------------------------------- Reporter: diener | Owner: Type: defect | Status: new Priority: major | Milestone: ET_2011_10 Component: EinsteinToolkit thorn | Version: development version Keywords: | -----------------------------------+---------------------------------------- When attempting to run the test suites on an IBM P7 machine using the IBM fortran and C compilers several source files didn't compile for different reasons.
In NullDecomp/src/NullDecomp_IO.F90 extra commas was present in some write statements. In the attached patch I have removed these commas.
In NullEvolve/src/NullEvol_DissipMask.F90 the min and max functions was used with arguments of different type. The constants 1.0 and 0.0 are by default single precision while CCTK_REAL are double precision. In the patch I have added CCTK_REAL parameters one and zero that are used throughout.
In SphericalHarmonicDecomp/src/h5read.cc and SphericalHarmonicRecon/src/Recon.cc the function sqrt was applied to variables of integer type and the compiler didn't know what to convert them to. In the attached patch I explicitly convert the integers to double before taking the sqrt.
#636: PITTNullCode does not compile on IBM P7 using the ibm compilers. ------------------------------------+--------------------------------------- Reporter: diener | Owner: Type: defect | Status: new Priority: major | Milestone: ET_2011_10 Component: EinsteinToolkit thorn | Version: development version Resolution: | Keywords: ------------------------------------+---------------------------------------
Comment (by hinder):
Why not convert the integers to CCTK_REAL rather than double?
In either case, the patch looks good. Please apply.
#636: PITTNullCode does not compile on IBM P7 using the ibm compilers. ------------------------------------+--------------------------------------- Reporter: diener | Owner: Type: defect | Status: new Priority: major | Milestone: ET_2011_10 Component: EinsteinToolkit thorn | Version: development version Resolution: | Keywords: ------------------------------------+---------------------------------------
Comment (by anonymous):
Replying to [comment:1 hinder]:
Why not convert the integers to CCTK_REAL rather than double?
In either case, the patch looks good. Please apply.
There was already other places in the code where converting to double was used. This can be cleaned up later.
#636: PITTNullCode does not compile on IBM P7 using the ibm compilers. ------------------------------------+--------------------------------------- Reporter: diener | Owner: Type: defect | Status: closed Priority: major | Milestone: ET_2011_10 Component: EinsteinToolkit thorn | Version: development version Resolution: fixed | Keywords: ------------------------------------+--------------------------------------- Changes (by diener):
* status: new => closed * resolution: => fixed
trac@lists.einsteintoolkit.org