Changes between Version 59 and Version 60 of NewC13

2015-07-24T17:42:47Z (2 years ago)
Gary J. Ferland

clean up comments about LLVM


  • NewC13

    v59 v60  
    1717Correct '''''fine optical depth calculation'''''.  Now uses the opacity for the current zone, rather than that from the previous zone calculated.  The major effect of this bug was in the transmitted continuum predicted close to the front face of optically-thick calculations, for the second and later iterations.  Reported by Jonathan Stern.
    19 The compilation will now work correctly on systems that '''''only have clang++ installed''''', and not g++. Both Mac Darwin and FreeBSD are moving the direction of dropping g++ in favor of clang++. On such systems you can do the compilation either in the main source directory, or the sys_llvm subdirectory.
     19The compilation will now work more or less correctly on systems that '''''only have LLVM (clang++) installed''''', and not GNU (g++). Both Mac Darwin and FreeBSD are moving in the direction of dropping g++ in favor of clang++.
     20LLVM does not allow trapping floating point errors, so we recommend that you use another compiler.
    2122From this version onwards we will '''''no longer trap FPEs when using the LLVM (clang++) compiler'''''. This makes debugging the code harder, but is necessary to get Cloudy to run with recent versions of Xcode.
     23We do not recommend using LLVM for this reason.
    2325We fixed a bug where Cloudy would not compile when using the '''''Intel MPI implementation'''''. Reported by Alex Arth.
    3133A bug has been fixed where '''''line emissivities got stuck at a small value''''' (e.g. in the output of SAVE LINES EMISSIVITY), when they should have decreased to zero.  Reported by Janet Simpson.
     35We have '''''changed our definition of the xi ionization parameter to be consistent with XSTAR'''''.
     36Through versions C13.03 we defined xi as the total luminosity in ionizing photons, the
     37original Tarter definition.  The X-ray community uses the Kallman & Bautista (2001) definition,
     38which integrates the SED between 1 and 1000 Rydbergs. 
     39The 1 - 1000 Ryd definition is part of C13.04 and later versions of the code.
     40This only makes a difference for very hard SEDs.
     41Reported by
     42Ildar Khabibullin.