V1.4 bugs in some Fortran compiler in Hitachi SR16000

[GoogleCodeExporter]

What steps will reproduce the problem?
Reported by T. Koshiro

> Now I'm trying to implement COSP v1.4 into our model, MRI-CGCM3. I'd
> like to report some errors I got in my environment. I hope it helps
> improve the next release.
> 
> 1) My F90 compiler (Optimizing FORTRAN90 compiler on Hitachi SR16000)
>    complains when a variable name exceeds 31 characters. So I changed
>    the variable names as follows:
> 
>    nominalPressureHistogramBoundaries -> 
> nominalPresHistogramBoundaries
>    Optical_Thickness_Total_MeanLog10 -> Opt_Thickness_Total_MeanLog10
>    Optical_Thickness_Water_MeanLog10 -> Opt_Thickness_Water_MeanLog10
>    Optical_Thickness_vs_Cloud_Top_Pressure -> 
> Opt_Thickness_vs_Cloud_Top_Pres
>    (in MODIS_simulator/modis_simulator.F90, cosp_modis_simulator.F90,
>     cosp_simulator.F90)
> 
>    RADAR_SIM_MICROPHYSICS_SCHEME_NAME -> 
> RADAR_SIM_MICROPHYS_SCHEME_NAME
>    RADAR_SIM_UPDATE_scale_LUTs_flag -> RADAR_SIM_UPDAT_scale_LUTs_flag
>    (in cosp_constants.F90, cosp_types.F90)
> 
> 
> 2) I got compile errors when the line feed characters of the source
>    file are CR+LF. So I changed them to LF in the following files
>    (This may not be exhaustive; all files had better to be checked,
>    just in case):
> 
>    cosp_types.F90
>    cosp_utils.F90
>    llnl/cosp_radar.F90
>    quickbeam/format_input.f90
> 
> 
> 3) For my compiler, function calls cannot be nested. So I rewrote Line
>    485 in MODIS_simulator/modis_simulator.F90 as follows:
> 
>    integer, dimension(size(phase, 1)) :: count_cloudMask
> 
>    ...
> 
>    count_cloudMask = count(cloudMask, dim = 2)
>    Cloud_Top_Pressure_Total_Mean =                         &
>        sum(cloud_top_pressure, mask = cloudMask, dim = 2)  &
>        / max(1, count_cloudMask)
> 

Original issue reported on code.google.com by alejandr...@metoffice.gov.uk on 12 Mar 2014 at 3:42