[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[netCDF #FYY-183264]: netcdf4.1.1 with hdf5-1.8.4-patch1
- Subject: [netCDF #FYY-183264]: netcdf4.1.1 with hdf5-1.8.4-patch1
- Date: Wed, 16 Mar 2011 06:30:47 -0600
Violeta,
> unfortunately it does not work. Here are the output files for
> netcdf-4.1.1
>
>
>
> I amd using szip-2.1, zlib-1.2.5, hdf5-1.8.5-patch1, netcdf-4.1.1
The error message appears to indicate the HDF5-1.8.6 is being used, rather than
1.8.5-patch1:
> *** Tests successful!
> PASS: tst_h_atts3
> HDF5-DIAG: Error detected in HDF5 (1.8.6) thread 0:
> #000: H5Ddeprec.c line 169 in H5Dcreate1(): unable to create dataset
> major: Dataset
> minor: Unable to initialize object
Is it possible there are multiple installations of HDF5, and you are linking to
the HDF5-1.8.6
library? We have reported a problem with the combination of netCDF-4.1.1 and
HDF5-1.8.6
as well as possible workarounds here:
http://www.unidata.ucar.edu/software/netcdf/docs/known_problems.html#incompat-411-186
We don't have ifort available on our MacOS X test platform, so can't duplicate
the problem you
are seeing, but our daily snapshot passes all tests using the cc compiler
that's part of the Xcode
developer tools with gfortran or g95.
> For HDF5 configure:
> ./configure --prefix=/usr/local/hdf5 --disable-shared --enable-
> fortran --with-zlib=/usr/local --with-szlib=/usr/local/szip
We don't use "--enable-fortran" with our HDF5 build, because the netCDF-4
library makes no use
of the HDF5 Fortran API. Instead the netCDF Fortran API is implemented on top
of the netCDF C library.
But I doubt if that is relevant to the problem you are seeing.
> For netcdf4 configure:
> ./configure --prefix=/usr/local/hdf5 --disable-shared --enable-
> fortran --with-zlib=/usr/local --with-szlib=/usr/local/szip /
> --enable-netcdf4 --enable-fortran --with-zlib=/usr/local --with-
> szlib=/usr/local/szip --with-hdf5=/usr/loca/hdf5 /
> --enable-separate-fortran
>
> My .profile:
> PATH=$PATH:$HOME:/usr/bin:/usr/local:/Applications/itt/idl/idl82/bin:/
> Applications/TeX:/opt/local/bin:~/scripts
> export CC=gcc
> export F9X=ifort
> export CXX=g++
> export FC=ifort
> export F77=ifort
> export F90=ifort
> export HDF5_DIR=/usr/local/hdf5
> export HDF5_LIB_DIR=/usr/local/lib
> export HDF5DIR=/usr/local/hdf5
> export NETCDF4_DIR=/usr/local/netcdf4
> export FCFLAGS='-O3'
> export FFLAGS='-O3'
> export CFLAGS='-O3'
> export CXXFLAGS='-O3 '
> export MACOSX_DEPLOYMENT_TARGET=10.5
> export LDFLAGS='-O3 '
> export DYLD_LIBRARY_PATH=/usr/local/lib:/usr/local/hdf5/:/usr/local/
> netcdf4:/usr/local/szip/lib:${DYLD_LIBRARY_PATH}
>
> the ifort is /opt/intel/Compiler/11.0/054/bin/ia32/ifort
>
> I did try with different compile flags like '-O3 -xN -ip -no-prec-div
> -static' as per Intel advice but no change.
We do have successful tests of our daily snapshot with ifort 10.1 on a linux
platform,
using no extra flags, just
CC=icc CFLAGS="-g -O2" FC=ifort FCFLAGS=-g F77=ifort FFLAGS=-g
--Russ
> On Mar 15, 2011, at 4:58 PM, Unidata netCDF Support wrote:
>
> > Hi Violeta,
> >
> >> I am tryind to install netsdf4 on and Mac OS X 10.5.8 with Intel
> >> FORTRAN compiles ifort version 11.0.
> >>
> >> The "make check" fails 1 out of 59 tests. Here it is attached the
> >> make.out.txt t file with the output.
> > ...
> >> *** testing long string variable...ok.
> >> *** Tests successful!
> >> PASS: tst_strings
> >> HDF5-DIAG: Error detected in HDF5 (1.8.4-patch1) thread 0:
> >> #000: H5O.c line 676 in H5Oget_info_by_idx(): group not found
> >> major: Symbol table
> >> minor: Object not found
> >> #001: H5Gloc.c line 578 in H5G_loc_find_by_idx(): can't find object
> >> major: Symbol table
> >> minor: Object not found
> > ...
> >> ================================================
> >> 1 of 59 tests failed
> >> Please report to address@hidden
> >> ================================================
> >> make[2]: *** [check-TESTS] Error 1
> >
> > This looks like a problem that was fixed in HDF5-1.8.5-patch1.
> > Could you
> > please try installing that version instead of HDF5-1.8.4-patch1 and
> > let
> > us know if the problem persists? Note, please don't use the later
> > release
> > 1.8.6 of HDF5, which has an incompatibility with netCDF 4.1.1 that
> > we don't
> > have a fix for yet.
> >
> > I hope this works.
> >
> > --Russ
> >
> >
> >
> >
> > Russ Rew UCAR Unidata Program
> > address@hidden http://www.unidata.ucar.edu
> >
> >
> >
> > Ticket Details
> > ===================
> > Ticket ID: FYY-183264
> > Department: Support netCDF
> > Priority: Normal
> > Status: Closed
> >
> >
> >
>
>
>
Russ Rew UCAR Unidata Program
address@hidden http://www.unidata.ucar.edu
Ticket Details
===================
Ticket ID: FYY-183264
Department: Support netCDF
Priority: Normal
Status: Closed