Hi, > This isn't really a support request, just a bit of additional > information that I wanted to send to you. Thanks! I've included your report in our "other builds" page at http://www.unidata.ucar.edu/software/netcdf/docs/other-builds.html Maybe it will be of help to future netCDF users. --Russ > I am using NetCDF 4.1.2 compiled with the Solaris Studio 12 update 2 > compilers (the new name for Sun Studio) on Linux (CentOS release 5.5), > which is one of your "other builds". Another user reported a working > setup previously, with the combination NetCDF 4.1 / SS 12.1. The > resulting code mostly works but fails the make check because of some > NaN issues ("NetCDF: Numeric conversion not representable"). > > This user sets "-fast" in his FLAGS, which is a macro that includes > "-fsimple=2", the aggressive floating point optimization flag. The SS > 12.2 documentation states (amongst other things) that: > * -fsimple=0 is the only one preserving strict IEEE 754 conformance > * -fsimple=1 assumes that "computations with Infinity or NaNs (?Not > a Number?) as operands need not propagate NaNs to their results" > > "-fsimple=1" doesn't work (checked), so the solution here is to > specify "-fsimple=0" after "-fast" (which should work but I didn't > test it!), or to set things up like this: > > CPPFLAGS="-DsunFortran" > CC=suncc CFLAGS="-O3 -xtarget=native -fsimple=0" > CXX=sunCC CXXFLAGS="-O3 -xtarget=native -fsimple=0" > F77=sunf77 FFLAGS="-O3 -xtarget=native -fsimple=0 -dalign" > FC=sunf90 FCFLAGS="-O3 -xtarget=native -fsimple=0 -dalign" > > Note that adding "-D__FUNCTION__=__func__" to the CPPFLAGS isn't > necessary anymore. > > I also needed the attached patch to "configure" as for some reason > autoconf is broken on the development machine here, so impossible to > do autoreconf. > > The configuration line I used was: > ./configure --prefix=${PREFIXDIR} \ > --enable-static \ > --enable-shared \ > --enable-large-file-tests \ > --with-hdf5=${PREFIXDIR} \ > --with-szlib=${PREFIXDIR} \ > --with-libcf \ > --enable-netcdf4 \ > --enable-cxx-4 > > With all that, all tests pass and further use of the library didn't > show any error. Note that enabling netdcf4 without enabling cxx-4 will > result in a build error. > > Thanks! > Regards, > JF Le FillÃtre > > -- > Departamento de IngenierÃa MatemÃtica > Facultad de Ciencias FÃsicas y MatemÃticas > Universidad de Chile Russ Rew UCAR Unidata Program address@hidden http://www.unidata.ucar.edu Ticket Details =================== Ticket ID: XWL-678268 Department: Support netCDF Priority: Normal Status: Closed
NOTE: All email exchanges with Unidata User Support are recorded in the Unidata inquiry tracking system and then made publicly available through the web. If you do not want to have your interactions made available in this way, you must let us know in each email you send to us.