Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Issue building ocprint with GCC 10.1...maybe? #1725

Open
mathomp4 opened this issue May 12, 2020 · 11 comments
Open

Issue building ocprint with GCC 10.1...maybe? #1725

mathomp4 opened this issue May 12, 2020 · 11 comments

Comments

@mathomp4
Copy link

Prelims

  • the version of the software with which you are encountering an issue
    • netCDF-C 4.7.4
  • environmental information (i.e. Operating System, compiler info, java version, python version, etc.)
    • macOS 10.15.4
    • GCC 9.3.0 and 10.1.0
    • Open MPI 4.0.3
  • a description of the issue with the steps needed to reproduce it
    • see below

The Issue

This is an odd one, mainly because this might be the second or third time ocprint and I have had issues. To wit, I recently built GCC 10.1.0 and am trying to build netCDF-c with it. Now, I have also built (and usually use) GCC 9.3.0 and with it, I can build netCDF-C just fine and ocprint seems to build. I don't know what it does, but it builds.

In each case, I am compiling using Open MPI wrappers as CC, CXX, and FC as I built HDF5 with parallel support. Each compiler is using Open MPI 4.0.3.

In each case, netcdf were configured within a framework as, essentially:

  $ ./configure --prefix=/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin --includedir=/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/netcdf --enable-hdf4 --disable-dap --enable-parallel-tests --disable-shared --disable-examples --enable-netcdf-4 CC=mpicc FC=mpifort CXX=mpic++ F77=mpifort

This is actually a lot of FC=$(FC) and --prefix=$(prefix), etc. in my overarching Base Library framework make system, but the configure line for each is nigh-identical save for the changes in prefix/includedir.

I've never seen this error with GCC 9.3.0 (though I'm not saying I can't), but I occasionally see it with GCC 10.1.0...and it seems weirdly non-deterministic.

First, with GCC 9.3.0. I search the log for nc4print.o and get to:

depbase=`echo nc4print.o | sed 's|[^/]*$|.deps/&|;s|\.o$||'`;\
mpicc -DHAVE_CONFIG_H -I. -I..  -I../include -I../oc2 -I/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/    -I/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/zlib    -I/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/szlib    -I/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/jpeg    -I/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/hdf5    -I/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/hdf    -I/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/netcdf    -I/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/udunits2    -I/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/gsl    -I/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/antlr     -MT nc4print.o -MD -MP -MF $depbase.Tpo -c -o nc4print.o nc4print.c &&\
mv -f $depbase.Tpo $depbase.Po
/bin/sh ../libtool  --tag=CC   --mode=link mpicc     -o nc4print nc4print.o ../liblib/libnetcdf.la -ljpeg -lmfhdf -ldf -ljpeg -lhdf5_hl -lhdf5 -lm -L/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib -lmfhdf -ldf -lsz -ljpeg  -L/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib -lcurl -lz -lm        -ldl -lm
libtool: link: mpicc -o nc4print nc4print.o  ../liblib/.libs/libnetcdf.a -L/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib /Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libhdf5_hl.a /Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libhdf5.a /Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libmfhdf.a /Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libdf.a /Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libsz.a /Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libjpeg.a /Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libcurl.a -lz -ldl -lm
depbase=`echo ocprint.o | sed 's|[^/]*$|.deps/&|;s|\.o$||'`;\
mpicc -DHAVE_CONFIG_H -I. -I..  -I../include -I../oc2 -I/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/    -I/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/zlib    -I/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/szlib    -I/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/jpeg    -I/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/hdf5    -I/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/hdf    -I/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/netcdf    -I/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/udunits2    -I/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/gsl    -I/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/antlr     -MT ocprint.o -MD -MP -MF $depbase.Tpo -c -o ocprint.o ocprint.c &&\
mv -f $depbase.Tpo $depbase.Po
/bin/sh ../libtool  --tag=CC   --mode=link mpicc     -o ocprint ocprint.o ../liblib/libnetcdf.la -ljpeg -lmfhdf -ldf -ljpeg -lhdf5_hl -lhdf5 -lm -L/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib -lmfhdf -ldf -lsz -ljpeg  -L/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib -lcurl -lz -lm  -ldl -lm
libtool: link: mpicc -o ocprint ocprint.o  ../liblib/.libs/libnetcdf.a -L/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib /Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libhdf5_hl.a /Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libhdf5.a /Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libmfhdf.a /Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libdf.a /Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libsz.a /Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libjpeg.a /Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libcurl.a -lz -ldl -lm
make[5]: Entering directory '/Users/mathomp4/Baselibs/ESMA-Baselibs-6.0.12-Snap13/src/netcdf/ncdump'
 /Users/mathomp4/.homebrew/brew/bin/gmkdir -p '/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/bin'
  /bin/sh ../libtool   --mode=install /Users/mathomp4/.homebrew/brew/bin/ginstall -c ncdump nccopy '/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/bin'
libtool: install: /Users/mathomp4/.homebrew/brew/bin/ginstall -c ncdump /Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/bin/ncdump
libtool: install: /Users/mathomp4/.homebrew/brew/bin/ginstall -c nccopy /Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/bin/nccopy
 /Users/mathomp4/.homebrew/brew/bin/gmkdir -p '/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/share/man/man1'
 /Users/mathomp4/.homebrew/brew/bin/ginstall -c -m 644 ncdump.1 nccopy.1 '/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/share/man/man1'
make[5]: Leaving directory '/Users/mathomp4/Baselibs/ESMA-Baselibs-6.0.12-Snap13/src/netcdf/ncdump'
make[4]: Leaving directory '/Users/mathomp4/Baselibs/ESMA-Baselibs-6.0.12-Snap13/src/netcdf/ncdump'
make[3]: Leaving directory '/Users/mathomp4/Baselibs/ESMA-Baselibs-6.0.12-Snap13/src/netcdf/ncdump'
Making install in unit_test

You can see we build nc4print then ocprint and then eventually install ncdump and nccopy.

Now, sometimes with netCDF-C, using GCC 10.1, I get:

depbase=`echo nc4print.o | sed 's|[^/]*$|.deps/&|;s|\.o$||'`;\
mpicc -DHAVE_CONFIG_H -I. -I..  -I../include  -I/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/    -I/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/zlib    -I/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/szlib    -I/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/jpeg    -I/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/hdf5    -I/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/hdf    -I/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/netcdf    -I/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/udunits2    -I/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/gsl    -I/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/antlr     -MT nc4print.o -MD -MP -MF $depbase.Tpo -c -o nc4print.o nc4print.c &&\
mv -f $depbase.Tpo $depbase.Po
/bin/sh ../libtool  --tag=CC   --mode=link mpicc     -o nc4print nc4print.o ../liblib/libnetcdf.la -ljpeg -lmfhdf -ldf -ljpeg -lhdf5_hl -lhdf5 -lm -lz -L/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib -lmfhdf -ldf -lsz -ljpeg   -ldl -lm
libtool: link: mpicc -o nc4print nc4print.o  ../liblib/.libs/libnetcdf.a -L/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libhdf5_hl.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libhdf5.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libmfhdf.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libdf.a -lz /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libsz.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libjpeg.a -ldl -lm
make[5]: Entering directory '/Users/mathomp4/Baselibs/ESMA-Baselibs-6.0.12-Snap13/src/netcdf/ncdump'
 /Users/mathomp4/.homebrew/brew/bin/gmkdir -p '/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/bin'
  /bin/sh ../libtool   --mode=install /Users/mathomp4/.homebrew/brew/bin/ginstall -c ncdump nccopy '/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/bin'
libtool: install: /Users/mathomp4/.homebrew/brew/bin/ginstall -c ncdump /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/bin/ncdump
libtool: install: /Users/mathomp4/.homebrew/brew/bin/ginstall -c nccopy /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/bin/nccopy
 /Users/mathomp4/.homebrew/brew/bin/gmkdir -p '/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/share/man/man1'
 /Users/mathomp4/.homebrew/brew/bin/ginstall -c -m 644 ncdump.1 nccopy.1 '/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/share/man/man1'
make[5]: Leaving directory '/Users/mathomp4/Baselibs/ESMA-Baselibs-6.0.12-Snap13/src/netcdf/ncdump'
make[4]: Leaving directory '/Users/mathomp4/Baselibs/ESMA-Baselibs-6.0.12-Snap13/src/netcdf/ncdump'
make[3]: Leaving directory '/Users/mathomp4/Baselibs/ESMA-Baselibs-6.0.12-Snap13/src/netcdf/ncdump'
Making install in unit_test

Once again we make nc4print and then...skip over ocprint and go right to installing ncdump and nccopy! Thus, we avoid my possible issue. I've had this happen both with a netcdf where I've done make clean && make distclean and when I've done a complete new clone/checkout of netcdf (or my full base libraries).

But, other times, I compile with GCC 10.1 and I get:

depbase=`echo nc4print.o | sed 's|[^/]*$|.deps/&|;s|\.o$||'`;\
mpicc -DHAVE_CONFIG_H -I. -I..  -I../include -I../oc2 -I/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/    -I/Users/mathomp4/installed/
MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/zlib    -I/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/sz
lib    -I/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/jpeg    -I/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Basel
ibs/6.0.12-Snap13/Darwin/include/hdf5    -I/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/hdf    -I/Users/mathomp4/installed/MPI/gcc-gf
ortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/netcdf    -I/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/udunits2
  -I/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/gsl    -I/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.
0.12-Snap13/Darwin/include/antlr     -MT nc4print.o -MD -MP -MF $depbase.Tpo -c -o nc4print.o nc4print.c &&\
mv -f $depbase.Tpo $depbase.Po
depbase=`echo ocprint.o | sed 's|[^/]*$|.deps/&|;s|\.o$||'`;\
mpicc -DHAVE_CONFIG_H -I. -I..  -I../include -I../oc2 -I/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/    -I/Users/mathomp4/installed/
MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/zlib    -I/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/sz
lib    -I/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/jpeg    -I/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Basel
ibs/6.0.12-Snap13/Darwin/include/hdf5    -I/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/hdf    -I/Users/mathomp4/installed/MPI/gcc-gf
ortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/netcdf    -I/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/udunits2
  -I/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/gsl    -I/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/include/antlr     -MT ocprint.o -MD -MP -MF $depbase.Tpo -c -o ocprint.o ocprint.c &&\
mv -f $depbase.Tpo $depbase.Po
/bin/sh ../libtool  --tag=CC   --mode=link mpicc     -o ncdump ncdump.o vardata.o dumplib.o indent.o nctime0.o utils.o nciter.o ../liblib/libnetcdf.la -ljpeg -lmfhdf -ldf -ljpeg -lhdf5_hl -lhdf5 -lm -L/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib -lmfhdf -ldf -lsz -ljpeg  -L/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib -lcurl -lz -lm  -ldl -lm
/bin/sh ../libtool  --tag=CC   --mode=link mpicc     -o nccopy nccopy.o nciter.o chunkspec.o utils.o dimmap.o list.o ../liblib/libnetcdf.la -ljpeg -lmfhdf -ldf -ljpeg -lhdf5_hl -lhdf5 -lm -L/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib -lmfhdf -ldf -lsz -ljpeg  -L/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib -lcurl -lz -lm     -ldl -lm
/bin/sh ../libtool  --tag=CC   --mode=link mpicc     -o ncvalidator ncvalidator.o ../liblib/libnetcdf.la -ljpeg -lmfhdf -ldf -ljpeg -lhdf5_hl -lhdf5 -lm -L/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib -lmfhdf -ldf -lsz -ljpeg  -L/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib -lcurl -lz -lm        -ldl -lm
libtool: link: mpicc -o ncdump ncdump.o vardata.o dumplib.o indent.o nctime0.o utils.o nciter.o  ../liblib/.libs/libnetcdf.a -L/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libhdf5_hl.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libhdf5.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libmfhdf.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libdf.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libsz.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libjpeg.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libcurl.a -lz -ldl -lm
libtool: link: mpicc -o nccopy nccopy.o nciter.o chunkspec.o utils.o dimmap.o list.o  ../liblib/.libs/libnetcdf.a -L/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libhdf5_hl.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libhdf5.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libmfhdf.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libdf.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libsz.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libjpeg.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libcurl.a -lz -ldl -lm
libtool: link: mpicc -o ncvalidator ncvalidator.o  ../liblib/.libs/libnetcdf.a -L/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libhdf5_hl.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libhdf5.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libmfhdf.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libdf.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libsz.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libjpeg.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libcurl.a -lz -ldl -lm
/bin/sh ../libtool  --tag=CC   --mode=link mpicc     -o nc4print nc4print.o ../liblib/libnetcdf.la -ljpeg -lmfhdf -ldf -ljpeg -lhdf5_hl -lhdf5 -lm -L/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib -lmfhdf -ldf -lsz -ljpeg  -L/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib -lcurl -lz -lm      -ldl -lm
/bin/sh ../libtool  --tag=CC   --mode=link mpicc     -o ocprint ocprint.o ../liblib/libnetcdf.la -ljpeg -lmfhdf -ldf -ljpeg -lhdf5_hl -lhdf5 -lm -L/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib -lmfhdf -ldf -lsz -ljpeg  -L/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib -lcurl -lz -lm        -ldl -lm
libtool: link: mpicc -o nc4print nc4print.o  ../liblib/.libs/libnetcdf.a -L/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libhdf5_hl.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libhdf5.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libmfhdf.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libdf.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libsz.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libjpeg.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libcurl.a -lz -ldl -lm
libtool: link: mpicc -o ocprint ocprint.o  ../liblib/.libs/libnetcdf.a -L/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libhdf5_hl.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libhdf5.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libmfhdf.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libdf.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libsz.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libjpeg.a /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib/libcurl.a -lz -ldl -lm
duplicate symbol '_ocdebug' in:
    ocprint.o
    ../liblib/.libs/libnetcdf.a(liboc_la-ocdebug.o)
ld: 1 duplicate symbol for architecture x86_64
collect2: error: ld returned 1 exit status
make[4]: *** [Makefile:1106: ocprint] Error 1
make[4]: *** Waiting for unfinished jobs....
make[4]: Leaving directory '/Users/mathomp4/Baselibs/ESMA-Baselibs-6.0.12-Snap13/src/netcdf/ncdump'
make[3]: *** [Makefile:1332: install-recursive] Error 1
make[3]: Leaving directory '/Users/mathomp4/Baselibs/ESMA-Baselibs-6.0.12-Snap13/src/netcdf/ncdump'
make[2]: *** [Makefile:707: install-recursive] Error 1

I am so confused. Perhaps make distclean is not enough to clean netCDF-C? Why do I sometimes get ocprint building and other times I don't?

I also thought this is perhaps a race condition as I usually build make -j3 or make -j8. But, I hacked my build scripts to force make -j1 install when building netcdf, so it should be serial.

Are there any thoughts on what I might be doing wrong? As I said before, I did have ocprint issues on macOS before, but 4.6.2 solved it for me and I've never had issues since.

@DennisHeimbigner
Copy link
Collaborator

First, WRT not building ocprint. This is conditional on enabling DAP.
This in turn requires access to libcurl. So my speculation is that
libcurl is not always being found. If you look at the ouput
from ./configure, it should have a line something like:

checking for curl_easy_setopt in -lcurl .. yes
If it says no, then libcurl was not found (or was misconfigured)
and DAP will be disabled.

@DennisHeimbigner
Copy link
Collaborator

Second, the duplicate symbol error is actually correct.
My surmise is that gcc 9 either recognizes that the two
declarations are the same, or issues a warning --
but since I do not see it in the output you sent, it is probably
the former. In any case, gcc 9 and gcc 10 are handling this
different.
I will issue a patch, but in the meantime, you can try the following:

  1. FInd the following line in ncdump/ocprint,c (about line 59)

int ocdebug;

  1. Change that line to this:

extern int ocdebug;

@mathomp4
Copy link
Author

First, WRT not building ocprint. This is conditional on enabling DAP.
This in turn requires access to libcurl. So my speculation is that
libcurl is not always being found. If you look at the ouput
from ./configure, it should have a line something like:

checking for curl_easy_setopt in -lcurl .. yes
If it says no, then libcurl was not found (or was misconfigured)
and DAP will be disabled.

@DennisHeimbigner Ah! Indeed. I do see it saying yes and no on and off in my log files, which isn't good.

As far as I know, curl is building with GCC 10.1 just fine. My GNUmakefile requires a good cURL build to proceed to netcdf (along with hdf5). And I can see it there. curl the executable is built and curl-config between GCC 9.3 and 10.1 are identical (save for directory names). Do you know if:

./configure --prefix=foo --includedir=foo --libdir=foo
 --with-zlib=/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin
 --disable-ldap --enable-manual 
--disable-shared --enable-static 
--without-libidn --without-libidn2 CFLAGS= CC=gcc CXX=g++ FC=gfortran

Is enough to build curl_easy_setopt or perhaps I need more/less? Very baffling...

DennisHeimbigner added a commit to DennisHeimbigner/netcdf-c that referenced this issue May 12, 2020
re: Issue Unidata#1725

The program ncdump/ocprint has a duplicate definition of
the variable ocdebug. Convert to an extern definition
since it is define in oc2 code.
@DennisHeimbigner
Copy link
Collaborator

Can you post the output of config.log for the case when
ocprint is not being compiled?

@DennisHeimbigner
Copy link
Collaborator

DennisHeimbigner commented May 12, 2020

Ah! Indeed. I do see it saying yes and no on and off in my log files, which isn't good.

So you are saying that for gcc 9, that sometimes curl_easy_setopt
is found and other times it is not? Is there any pattern connecting
this behavior to something else like a change in the
./configure options?

@mathomp4
Copy link
Author

@DennisHeimbigner

As far as I can tell, GCC 9 is always "curl good" if you were. I just looked at my logs and compared the curl configure output from my builds and the only substantive difference in how it configured was:

❯ diff curl.930.log curl.1010.log
31d30
< parallel-tests: installing './test-driver'
726c725
<   Install prefix:   /Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin
---
>   Install prefix:   /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin

I can't imagine "parallel-tests" would cause a fundamental difference.

What I usually do is if I see curl in my libraries, I assume we want to build with DAP support so I then do:

BUILD_DAP = --enable-dap
LIB_CURL = $(shell $(prefix)/bin/curl-config --libs)
netcdf.config : netcdf/configure
        @echo "Configuring netcdf $*"
        @(cd netcdf; \
          export PATH="$(prefix)/bin:$(PATH)" ;\
          export CPPFLAGS="$(CPPFLAGS) $(NC_CPPFLAGS) $(INC_SUPP)";\
          export CFLAGS="$(CFLAGS) $(NC_CFLAGS) $(PTHREAD_FLAG)";\
          export LIBS="-L$(prefix)/lib -lmfhdf -ldf -lsz -ljpeg $(LINK_GPFS) $(LIB_CURL) -ldl -lm $(LIB_EXTRA)" ;\
          ./configure --prefix=$(prefix) \
                      --includedir=$(prefix)/include/netcdf \
                      --enable-hdf4 \
                      $(BUILD_DAP) \
                      $(NC_PAR_TESTS) \
                      --disable-shared \
                      --disable-examples \
                      --enable-netcdf-4 \
                      CC=$(NC_CC) FC=$(NC_FC) CXX=$(NC_CXX) F77=$(NC_F77) )
        @touch $@

There's a lot of extra cruft that in many cases is blank nowadays, but I rely on curl-config to provide what it wants to netcdf. And for GCC 9:

❯ /Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/bin/curl-config --libs
-L/Users/mathomp4/installed/MPI/gcc-gfortran-9.3.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib -lcurl -lz -lm

and for GCC 10:

❯ /Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/bin/curl-config --libs
-L/Users/mathomp4/installed/MPI/gcc-gfortran-10.1.0/openmpi-4.0.3/Baselibs/6.0.12-Snap13/Darwin/lib -lcurl -lz -lm

Honestly, the curl-config is a bit overboard, but I use nc-config and nf-config so much (with GNUmake), if I see something like it, I use it.

@mathomp4
Copy link
Author

I think I might have to move over to cURL and see if they have ideas...

@DennisHeimbigner
Copy link
Collaborator

You probably do not have to do the --enable-dap option
because it is on by default unless it cannot find the curl library.
Using curl-config is ok; I do it myself.
At this point, I am baffled. There must be difference somewhere.
Can you post the following information:

  1. the exact ./configure command for the case where curl is found
    and the case where curl is not found.
  2. the config.log file for when curl is found and the file when curl is not found.

@mathomp4
Copy link
Author

Oooh. You might have twigged it... Something weird is happening in make land. For some reason my make verify command (which prints out a lot of the variables in my GNUmakefile) seems to say I'll be passing in --enable-dap but when it builds netcdf suddenly that is --disable-dap.

Time to make a much more bulletproof "am I building with curl" test...even though my current one has worked for at least 5 years. (Grah. GNU Make!)

@mathomp4
Copy link
Author

Well, if I hardcode --enable-dap I can have it fail consistently. Yay! (Yay?)

Then, if I try your change to ocprint.c in #1726, all works and ocprint is happy.

Thank you @DennisHeimbigner for figuring this out. (Next philosophical question: why is it always ocprint when I come here... 😄 )

Now time to figure out how to dynamically patch code in GNU Make since my base library repo just has netcdf as a submodule checkout out at v4.7.4...

@DennisHeimbigner
Copy link
Collaborator

I guess ocprint is your white whale :-)

DennisHeimbigner added a commit to DennisHeimbigner/netcdf-c that referenced this issue May 19, 2020
Fix Issue Unidata#1725.
Replace PR Unidata#1726
Also replace PR Unidata#1694

The general problem is that under Visual Studio, we are seeing
a number of undefined reference and other scoping errors.
The reason is that the code is not properly using Visual Studio
_declspec() declarations.

The basic solution is to ensure that when compiling the code itself
one needs to ensure that _declspec(dllexport) is used. There
are several sets of macros to handle this, but they all rely
on the flag DLL_EXPORT being define when the code is compiled,
but not being defined when the code is used via a .h file.

As a test, I modified XGetOpt.c to build properly. I also
fixed the oc2 library to properly _declspec things like ocdebug.

I also made some misc. changes to get all the tests to run
if cygwin is installed (to get bash, sed, etc).

Misc. Changes:
* Put XGetOpt.c into libsrc and copy at build time
  to the other directories where it is needed.
netbsd-srcmastr pushed a commit to NetBSD/pkgsrc that referenced this issue Jun 3, 2023
Release Notes       {#RELEASE_NOTES}
=============

\brief Release notes file for the netcdf-c package.

This file contains a high-level description of this package's
evolution. Releases are in reverse chronological order (most recent
first). Note that, as of netcdf 4.2, the `netcdf-c++` and
`netcdf-fortran` libraries have been separated into their own
libraries.

## 4.9.3 - TBD

## 4.9.2 - March 14, 2023

This is the maintenance release which adds support for HDF5 version
1.14.0, in addition to a handful of other changes and bugfixes.

* Fix 'make distcheck' error in run_interop.sh. See [Github
  #????](https://github.com/Unidata/netcdf-c/pull/????).
* Update `nc-config` to remove inclusion from automatically-detected
  `nf-config` and `ncxx-config` files, as the wrong files could be
  included in the output.  This is in support of [GitHub
  #2274](Unidata/netcdf-c#2274).

* Update H5FDhttp.[ch] to work with HDF5 version 1.13.2 and later. See
  [Github #2635](Unidata/netcdf-c#2635).

* [Bug Fix] Update DAP code to enable CURLOPT_ACCEPT_ENCODING by
  default. See [Github
  #2630](Unidata/netcdf-c#2630).

* [Bug Fix] Fix byterange failures for certain URLs. See [Github
  #2649](Unidata/netcdf-c#2649).

* [Bug Fix] Fix 'make distcheck' error in run_interop.sh. See [Github
  #2631](Unidata/netcdf-c#2631).

* [Enhancement] Update `nc-config` to remove inclusion from
  automatically-detected `nf-config` and `ncxx-config` files, as the
  wrong files could be included in the output.  This is in support of
  [GitHub #2274](Unidata/netcdf-c#2274).

* [Enhancement] Update H5FDhttp.[ch] to work with HDF5 version
  1.14.0. See [Github
  #2615](Unidata/netcdf-c#2615).

## 4.9.1 - February 2, 2023

## Known Issues

* A test in the `main` branch of `netcdf-cxx4` is broken by this rc; this will
  bear further investigation, but not being treated as a roadblock for the
  release candidate.

* The new document, `netcdf-c/docs/filter_quickstart.md` is in
  rough-draft form.

* Race conditions exist in some of the tests when run concurrently with large
  numbers of processors

## What's Changed from v4.9.0 (automatically generated)

* Fix nc_def_var_fletcher32 operation by \@DennisHeimbigner in
  Unidata/netcdf-c#2403

* Merge relevant info updates back into `main` by \@WardF in
  Unidata/netcdf-c#2387

* Add manual GitHub actions triggers for the tests. by \@WardF in
  Unidata/netcdf-c#2404

* Use env variable USERPROFILE instead of HOME for windows and mingw. by
  \@DennisHeimbigner in Unidata/netcdf-c#2405

* Make public a limited API for programmatic access to internal .rc tables by
  \@DennisHeimbigner in Unidata/netcdf-c#2408

* Fix typo in CMakeLists.txt by \@georgthegreat in
  Unidata/netcdf-c#2412

* Fix choice of HOME dir by \@DennisHeimbigner in
  Unidata/netcdf-c#2416

* Check for libxml2 development files by \@WardF in
  Unidata/netcdf-c#2417

* Updating Doxyfile.in with doxygen-1.8.17, turned on WARN_AS_ERROR, added
  doxygen build to CI run by \@edwardhartnett in
  Unidata/netcdf-c#2377

* updated release notes by \@edwardhartnett in
  Unidata/netcdf-c#2392

* increase read block size from 1 KB to 4 MB by \@wkliao in
  Unidata/netcdf-c#2319

* fixed RELEASE_NOTES.md by \@edwardhartnett in
  Unidata/netcdf-c#2423

* Fix pnetcdf tests in cmake by \@WardF in
  Unidata/netcdf-c#2437

* Updated CMakeLists to avoid corner case cmake error by \@WardF in
  Unidata/netcdf-c#2438

* Add `--disable-quantize` to configure by \@WardF in
  Unidata/netcdf-c#2439

* Fix the way CMake handles -DPLUGIN_INSTALL_DIR by \@DennisHeimbigner in
  Unidata/netcdf-c#2430

* fix and test quantize mode for NC_CLASSIC_MODEL by \@edwardhartnett in
  Unidata/netcdf-c#2445

* Guard _declspec(dllexport) in support of #2446 by \@WardF in
  Unidata/netcdf-c#2460

* Ensure that netcdf_json.h does not interfere with ncjson. by
  \@DennisHeimbigner in Unidata/netcdf-c#2448

* Prevent cmake writing to source dir by \@magnusuMET in
  Unidata/netcdf-c#2463

* more quantize testing and adding pre-processor constant NC_MAX_FILENAME to
  nc_tests.h by \@edwardhartnett in
  Unidata/netcdf-c#2457

* Provide a default enum const when fill value does not match any enum
  constant by \@DennisHeimbigner in
  Unidata/netcdf-c#2462

* Fix support for reading arrays of HDF5 fixed size strings by
  \@DennisHeimbigner in Unidata/netcdf-c#2466

* fix musl build by \@magnusuMET in
  Unidata/netcdf-c#1701

* Fix AWS SDK linking errors by \@dzenanz in
  Unidata/netcdf-c#2470

* Address jump-misses-init issue. by \@WardF in
  Unidata/netcdf-c#2488

* Remove stray merge conflict markers by \@WardF in
  Unidata/netcdf-c#2493

* Add support for Zarr string type to NCZarr by \@DennisHeimbigner in
  Unidata/netcdf-c#2492

* Fix some problems with PR 2492 by \@DennisHeimbigner in
  Unidata/netcdf-c#2497

* Fix some bugs in the blosc filter wrapper by \@DennisHeimbigner in
  Unidata/netcdf-c#2461

* Add option to control accessing external servers by \@DennisHeimbigner in
  Unidata/netcdf-c#2491

* Changed attribute case in documentation by \@WardF in
  Unidata/netcdf-c#2482

* Adding all-error-codes.md back in to distribution documentation. by \@WardF in
  Unidata/netcdf-c#2501

* Update hdf5 version in github actions. by \@WardF in
  Unidata/netcdf-c#2504

* Minor update to doxygen function documentation by \@gsjaardema in
  Unidata/netcdf-c#2451

* Fix some addtional errors in NCZarr by \@DennisHeimbigner in
  Unidata/netcdf-c#2503

* Cleanup szip handling some more by \@DennisHeimbigner in
  Unidata/netcdf-c#2421

* Check for zstd development headers in autotools by \@WardF in
  Unidata/netcdf-c#2507

* Add new options to nc-config by \@WardF in
  Unidata/netcdf-c#2509

* Cleanup built test sources in nczarr_test by \@DennisHeimbigner in
  Unidata/netcdf-c#2508

* Fix inconsistency in netcdf_meta.h by \@WardF in
  Unidata/netcdf-c#2512

* Small fix in nc-config.in by \@WardF in
  Unidata/netcdf-c#2513

* For loop initial declarations are only allowed in C99 mode by \@gsjaardema in
  Unidata/netcdf-c#2517

* Fix some dependencies in tst_nccopy3 by \@WardF in
  Unidata/netcdf-c#2518

* Update plugins/Makefile.am by \@WardF in
  Unidata/netcdf-c#2519

* Fix prereqs in ncdump/tst_nccopy4 in order to avoid race conditions. by
  \@WardF in Unidata/netcdf-c#2520

* Move construction of VERSION file to end of the build by \@DennisHeimbigner in
  Unidata/netcdf-c#2527

* Add draft filter quickstart guide by \@WardF in
  Unidata/netcdf-c#2531

* Turn off extraneous debug output by \@DennisHeimbigner in
  Unidata/netcdf-c#2537

* typo fix by \@wkliao in Unidata/netcdf-c#2538

* replace 4194304 with READ_BLOCK_SIZE by \@wkliao in
  Unidata/netcdf-c#2539

* Rename variable to avoid function name conflict by \@ibaned in
  Unidata/netcdf-c#2550

* Add Cygwin CI and stop installing unwanted plugins by \@DWesl in
  Unidata/netcdf-c#2529

* Merge subset of v4.9.1 files back into main development branch by \@WardF in
  Unidata/netcdf-c#2530

* Add a Filter quickstart guide document by \@WardF in
  Unidata/netcdf-c#2524

* Fix race condition in ncdump (and other) tests. by \@DennisHeimbigner in
  Unidata/netcdf-c#2552

* Make dap4 reference dap instead of hard-wired to be disabled. by \@WardF in
  Unidata/netcdf-c#2553

* Suppress nczarr_test/tst_unknown filter test by \@DennisHeimbigner in
  Unidata/netcdf-c#2557

* Add fenceposting for HAVE_DECL_ISINF and HAVE_DECL_ISNAN by \@WardF in
  Unidata/netcdf-c#2559

* Add an old static file. by \@WardF in
  Unidata/netcdf-c#2575

* Fix infinite loop in file inferencing by \@DennisHeimbigner in
  Unidata/netcdf-c#2574

* Merge Wellspring back into development branch by \@WardF in
  Unidata/netcdf-c#2560

* Allow ncdump -t to handle variable length string attributes by \@srherbener in
  Unidata/netcdf-c#2584

* Fix an issue I introduced with make distcheck by \@WardF in
  Unidata/netcdf-c#2590

* make UDF0 not require NC_NETCDF4 by \@jedwards4b in
  Unidata/netcdf-c#2586

* Expose user-facing documentation related to byterange DAP functionality.  by
  \@WardF in Unidata/netcdf-c#2596

* Fix Memory Leak by \@DennisHeimbigner in
  Unidata/netcdf-c#2598

* CI: Change autotools CI build to out-of-tree build. by \@DWesl in
  Unidata/netcdf-c#2577

* Update github action configuration scripts. by \@WardF in
  Unidata/netcdf-c#2600

* Update the filter quickstart guide.  by \@WardF in
  Unidata/netcdf-c#2602

* Fix symbol export on Windows by \@WardF in
  Unidata/netcdf-c#2604

## New Contributors
* \@georgthegreat made their first contribution in Unidata/netcdf-c#2412
* \@dzenanz made their first contribution in Unidata/netcdf-c#2470
* \@DWesl made their first contribution in Unidata/netcdf-c#2529
* \@srherbener made their first contribution in Unidata/netcdf-c#2584
* \@jedwards4b made their first contribution in Unidata/netcdf-c#2586

**Full Changelog**: Unidata/netcdf-c@v4.9.0...v4.9.1

### 4.9.1 - Release Candidate 2 - November 21, 2022

#### Known Issues

* A test in the `main` branch of `netcdf-cxx4` is broken by this rc;
  this will bear further investigation, but not being treated as a
  roadblock for the release candidate.

* The new document, `netcdf-c/docs/filter_quickstart.md` is in rough-draft form.

#### Changes

* [Bug Fix] Fix a race condition when testing missing filters. See
  [Github #2557](Unidata/netcdf-c#2557).

* [Bug Fix] Fix some race conditions due to use of a common file in multiple
  shell scripts . See [Github
  #2552](Unidata/netcdf-c#2552).


### 4.9.1 - Release Candidate 1 - October 24, 2022

* [Enhancement][Documentation] Add Plugins Quick Start Guide.  See
  [GitHub #2524](Unidata/netcdf-c#2524) for
  more information.

* [Enhancement] Add new entries in `netcdf_meta.h`, `NC_HAS_BLOSC` and
  `NC_HAS_BZ2`. See [Github
  #2511](Unidata/netcdf-c#2511) and [Github
  #2512](Unidata/netcdf-c#2512) for more
  information.

* [Enhancement] Add new options to `nc-config`: `--has-multifilters`,
  `--has-stdfilters`, `--has-quantize`, `--plugindir`.  See [Github
  #2509](Unidata/netcdf-c#2509) for more
  information.

* [Bug Fix] Fix some errors detected in PR 2497. [PR
  #2497](Unidata/netcdf-c#2497) . See [Github
  #2503](Unidata/netcdf-c#2503).

* [Bug Fix] Split the remote tests into two parts: one for the
  remotetest server and one for all other external servers. Also add a
  configure option to enable the latter set. See [Github
  #2491](Unidata/netcdf-c#2491).

* [Bug Fix] Fix blosc plugin errors. See [Github
  #2461](Unidata/netcdf-c#2461).

* [Bug Fix] Fix support for reading arrays of HDF5 fixed size
  strings. See [Github
  #2466](Unidata/netcdf-c#2466).

* [Bug Fix] Fix some errors detected in [PR
  #2492](Unidata/netcdf-c#2492) . See [Github
  #2497](Unidata/netcdf-c#2497).

* [Enhancement] Add support for Zarr (fixed length) string type in
  nczarr. See [Github
  #2492](Unidata/netcdf-c#2492).

* [Bug Fix] Split the remote tests into two parts: one for the
  remotetest server and one for all other external servers. Also add a
  configure option to enable the latter set. See [Github
  #2491](Unidata/netcdf-c#2491).

* [Bug Fix] Fix support for reading arrays of HDF5 fixed size
  strings. See [Github
  #2462](Unidata/netcdf-c#2466).

* [Bug Fix] Provide a default enum const when fill value does not
  match any enum constant for the value zero. See [Github
  #2462](Unidata/netcdf-c#2462).

* [Bug Fix] Fix the json submodule symbol conflicts between libnetcdf
  and the plugin specific netcdf_json.h. See [Github
  #2448](Unidata/netcdf-c#2448).

* [Bug Fix] Fix quantize with CLASSIC_MODEL files. See [Github
  #2405](Unidata/netcdf-c#2445).

* [Enhancement] Add `--disable-quantize` option to `configure`.

* [Bug Fix] Fix CMakeLists.txt to handle all acceptable boolean values
  for -DPLUGIN_INSTALL_DIR. See [Github
  #2430](Unidata/netcdf-c#2430).

* [Bug Fix] Fix tst_vars3.c to use the proper szip flag. See [Github
  #2421](Unidata/netcdf-c#2421).

* [Enhancement] Provide a simple API to allow user access to the
  internal .rc file table: supports get/set/overwrite of entries of
  the form "key=value". See [Github
  #2408](Unidata/netcdf-c#2408).

* [Bug Fix] Use env variable USERPROFILE instead of HOME for windows
  and mingw. See [Github
  #2405](Unidata/netcdf-c#2405).

* [Bug Fix] Fix the nc_def_var_fletcher32 code in hdf5 to properly
  test value of the fletcher32 argument. See [Github
  #2403](Unidata/netcdf-c#2403).

## 4.9.0 - June 10, 2022

* [Enhancement] Add quantize functions nc_def_var_quantize() and
  nc_inq_var_quantize() to enable lossy compression. See [Github
  #1548](Unidata/netcdf-c#1548).

* [Enhancement] Add zstandard compression functions nc_def_var_zstandard() and
  nc_inq_var_zstandard(). See [Github
  #2173](Unidata/netcdf-c#2173).

* [Enhancement] Have netCDF-4 logging output one file per processor when used
  with parallel I/O. See [Github
  #1762](Unidata/netcdf-c#1762).

* [Enhancement] Improve filter installation process to avoid use of an extra
  shell script. See [Github
  #2348](Unidata/netcdf-c#2348).

* [Bug Fix] Get "make distcheck" to work See [Github
  #2343](Unidata/netcdf-c#2343).

* [Enhancement] Allow the read/write of JSON-valued Zarr attributes to allow for
  domain specific info such as used by GDAL/Zarr. See [Github
  #2278](Unidata/netcdf-c#2278).

* [Enhancement] Turn on the XArray convention for NCZarr files by
  default. WARNING, this means that the mode should explicitly specify "nczarr"
  or "zarr" even if "xarray" or "noxarray" is specified. See [Github
  #2257](Unidata/netcdf-c#2257).

* [Enhancement] Update the documentation to match the current filter
  capabilities See [Github
  #2249](Unidata/netcdf-c#2249).

* [Enhancement] Update the documentation to match the current filter
  capabilities. See [Github
  #2249](Unidata/netcdf-c#2249).

* [Enhancement] Support installation of pre-built standard filters into
  user-specified location. See [Github
  #2318](Unidata/netcdf-c#2318).

* [Enhancement] Improve filter support. More specifically (1) add
  nc_inq_filter_avail to check if a filter is available, (2) add the notion of
  standard filters, (3) cleanup szip support to fix interaction with NCZarr. See
  [Github #2245](Unidata/netcdf-c#2245).

* [Enhancement] Switch to tinyxml2 as the default xml parser implementation. See
  [Github #2170](Unidata/netcdf-c#2170).

* [Bug Fix] Require that the type of the variable in nc_def_var_filter is not
  variable length. See [Github
  #/2231](Unidata/netcdf-c#2231).

* [File Change] Apply HDF5 v1.8 format compatibility when writing to previous
  files, as well as when creating new files.  The superblock version remains at
  2 for newly created files.  Full backward read/write compatibility for
  netCDF-4 is maintained in all cases.  See [Github
  #2176](Unidata/netcdf-c#2176).

* [Enhancement] Add ability to set dataset alignment for netcdf-4/HDF5
  files. See [Github #2206](Unidata/netcdf-c#2206).

* [Bug Fix] Improve UTF8 support on windows so that it can use utf8
  natively. See [Github #2222](Unidata/netcdf-c#2222).

* [Enhancement] Add complete bitgroom support to NCZarr. See [Github
  #2197](Unidata/netcdf-c#2197).

* [Bug Fix] Clean up the handling of deeply nested VLEN types. Marks
  nc_free_vlen() and nc_free_string as deprecated in favor of
  ncaux_reclaim_data(). See [Github
  #2179](Unidata/netcdf-c#2179).

* [Bug Fix] Make sure that netcdf.h accurately defines the flags in the
  open/create mode flags. See [Github
  #2183](Unidata/netcdf-c#2183).

* [Enhancement] Improve support for msys2+mingw platform. See [Github
  #2171](Unidata/netcdf-c#2171).

* [Bug Fix] Clean up the various inter-test dependencies in ncdump for
  CMake. See [Github #2168](Unidata/netcdf-c#2168).

* [Bug Fix] Fix use of non-aws appliances. See [Github
  #2152](Unidata/netcdf-c#2152).

* [Enhancement] Added options to suppress the new behavior from [Github
  #2135](Unidata/netcdf-c#2135).  The options for
  `cmake` and `configure` are, respectively `-DENABLE_LIBXML2` and
  `--(enable/disable)-libxml2`. Both of these options defaul to 'on/enabled'.
  When disabled, the bundled `ezxml` XML interpreter is used regardless of
  whether `libxml2` is present on the system.

* [Enhancement] Support optional use of libxml2, otherwise default to ezxml. See
  [Github #2135](Unidata/netcdf-c#2135) -- H/T to
  [Egbert Eich](https://github.com/e4t).

* [Bug Fix] Fix several os related errors. See [Github
  #2138](Unidata/netcdf-c#2138).

* [Enhancement] Support byte-range reading of netcdf-3 files stored in private
  buckets in S3. See [Github
  #2134](Unidata/netcdf-c#2134)

* [Enhancement] Support Amazon S3 access for NCZarr. Also support use of the
  existing Amazon SDK credentials system. See [Github
  #2114](Unidata/netcdf-c#2114)

* [Bug Fix] Fix string allocation error in H5FDhttp.c. See [Github
  #2127](Unidata/netcdf-c#2127).

* [Bug Fix] Apply patches for ezxml and for selected oss-fuzz detected
  errors. See [Github #2125](Unidata/netcdf-c#2125).

* [Bug Fix] Ensure that internal Fortran APIs are always defined. See [Github
  #2098](Unidata/netcdf-c#2098).

* [Enhancement] Support filters for NCZarr. See [Github
  #2101](Unidata/netcdf-c#2101)

* [Bug Fix] Make PR 2075 long file name be idempotent. See [Github
  #2094](Unidata/netcdf-c#2094).


## 4.8.1 - August 18, 2021

* [Bug Fix] Fix multiple bugs in libnczarr. See [Github
  #2066](Unidata/netcdf-c#2066).

* [Enhancement] Support windows network paths (e.g. \\svc\...). See [Github
  #2065](Unidata/netcdf-c#2065).

* [Enhancement] Convert to a new representation of the NCZarr meta-data
  extensions: version 2. Read-only backward compatibility is provided. See
  [Github #2032](Unidata/netcdf-c#2032).

* [Bug Fix] Fix dimension_separator bug in libnczarr. See [Github
  #2035](Unidata/netcdf-c#2035).

* [Bug Fix] Fix bugs in libdap4. See [Github
  #2005](Unidata/netcdf-c#2005).

* [Bug Fix] Store NCZarr fillvalue as a singleton instead of a 1-element
  array. See [Github #2017](Unidata/netcdf-c#2017).

* [Bug Fixes] The netcdf-c library was incorrectly determining the scope of
  dimension; similar to the type scope problem. See [Github
  #2012](Unidata/netcdf-c#2012) for more information.

* [Bug Fix] Re-enable DAP2 authorization testing. See [Github
  #2011](Unidata/netcdf-c#2011).

* [Bug Fix] Fix bug with windows version of mkstemp that causes failure to
  create more than 26 temp files. See [Github
  #1998](Unidata/netcdf-c#1998).

* [Bug Fix] Fix ncdump bug when printing VLENs with basetype char. See [Github
  #1986](Unidata/netcdf-c#1986).

* [Bug Fixes] The netcdf-c library was incorrectly determining the scope of
  types referred to by nc_inq_type_equal. See [Github
  #1959](Unidata/netcdf-c#1959) for more information.

* [Bug Fix] Fix bug in use of XGetopt when building under Mingw. See [Github
  #2009](Unidata/netcdf-c#2009).

* [Enhancement] Improve the error reporting when attempting to use a filter for
  which no implementation can be found in HDF5_PLUGIN_PATH. See [Github
  #2000](Unidata/netcdf-c#2000) for more information.

* [Bug Fix] Fix `make distcheck` issue in `nczarr_test/` directory. See [Github
  #2007](Unidata/netcdf-c#2007).

* [Bug Fix] Fix bug in NCclosedir in dpathmgr.c. See [Github
  #2003](Unidata/netcdf-c#2003).

* [Bug Fix] Fix bug in ncdump that assumes that there is a relationship between
  the total number of dimensions and the max dimension id. See [Github
  #2004](Unidata/netcdf-c#2004).

* [Bug Fix] Fix bug in JSON processing of strings with embedded quotes. See
  [Github #1993](Unidata/netcdf-c#1993).

* [Enhancement] Add support for the new "dimension_separator" enhancement to
  Zarr v2. See [Github #1990](Unidata/netcdf-c#1990) for
  more information.

* [Bug Fix] Fix hack for handling failure of shell programs to properly handle
  escape characters. See [Github
  #1989](Unidata/netcdf-c#1989).

* [Bug Fix] Allow some primitive type names to be used as identifiers depending
  on the file format. See [Github
  #1984](Unidata/netcdf-c#1984).

* [Enhancement] Add support for reading/writing pure Zarr storage format that
  supports the XArray _ARRAY_DIMENSIONS attribute. See [Github
  #1952](Unidata/netcdf-c#1952) for more information.

* [Update] Updated version of bzip2 used in filter testing/functionality, in
  support of [Github #1969](Unidata/netcdf-c#1969).

* [Bug Fix] Corrected HDF5 version detection logic as described in [Github
  #1962](Unidata/netcdf-c#1962).

## 4.8.0 - March 30, 2021

* [Enhancement] Bump the NC_DISPATCH_VERSION from 2 to 3, and as a side effect,
  unify the definition of NC_DISPATCH_VERSION so it only needs to be defined in
  CMakeLists.txt and configure.ac. See [Github
  #1945](Unidata/netcdf-c#1945) for more information.

* [Enhancement] Provide better cross platform path name management. This
  converts paths for various platforms (e.g. Windows, MSYS, etc.) so that they
  are in the proper format for the executing platform. See [Github
  #1958](Unidata/netcdf-c#1958) for more information.

* [Bug Fixes] The nccopy program was treating -d0 as turning deflation on rather
  than interpreting it as "turn off deflation". See [Github
  #1944](Unidata/netcdf-c#1944) for more information.

* [Enhancement] Add support for storing NCZarr data in zip files. See [Github
  #1942](Unidata/netcdf-c#1942) for more information.

* [Bug Fixes] Make fillmismatch the default for DAP2 and DAP4; too many servers
  ignore this requirement.

* [Bug Fixes] Fix some memory leaks in NCZarr, fix a bug with long strides in
  NCZarr. See [Github #1913](Unidata/netcdf-c#1913) for
  more information.

* [Enhancement] Add some optimizations to NCZarr, dosome cleanup of code cruft,
  add some NCZarr test cases, add a performance test to NCZarr. See [Github
  #1908](Unidata/netcdf-c#1908) for more information.

* [Bug Fix] Implement a better chunk cache system for NCZarr. The cache now uses
  extendible hashing plus a linked list for provide a combination of
  expandibility, fast access, and LRU behavior. See [Github
  #1887](Unidata/netcdf-c#1887) for more information.

* [Enhancement] Provide .rc fields for S3 authentication: HTTP.S3.ACCESSID and
  HTTP.S3.SECRETKEY.

* [Enhancement] Give the client control over what parts of a DAP2 URL are URL
  encoded (i.e. %xx). This is to support the different decoding rules that
  servers apply to incoming URLS. See [Github
  #1884](Unidata/netcdf-c#1884) for more information.

* [Bug Fix] Fix incorrect time offsets from `ncdump -t`, in some cases when the
  time `units` attribute contains both a **non-zero** time-of-day, and a time
  zone suffix containing the letter "T", such as "UTC".  See [Github
  #1866](Unidata/netcdf-c#1866) for more information.

* [Bug Fix] Cleanup the NCZarr S3 build options. See [Github
  #1869](Unidata/netcdf-c#1869) for more information.

* [Bug Fix] Support aligned access for selected ARM processors.  See [Github
  #1871](Unidata/netcdf-c#1871) for more information.

* [Documentation] Migrated the documents in the NUG/ directory to the dedicated
  NUG repository found at https://github.com/Unidata/netcdf

* [Bug Fix] Revert the internal filter code to simplify it. From the user's
  point of view, the only visible change should be that (1) the functions that
  convert text to filter specs have had their signature reverted and renamed and
  have been moved to netcdf_aux.h, and (2) Some filter API functions now return
  NC_ENOFILTER when inquiry is made about some filter. Internally, the dispatch
  table has been modified to get rid of the complex structures.

* [Bug Fix] If the HDF5 byte-range Virtual File Driver is available )HDf5 1.10.6
  or later) then use it because it has better performance than the one currently
  built into the netcdf library.

* [Bug Fix] Fixed byte-range support with cURL > 7.69. See
  [Unidata/netcdf-c#1798].

* [Enhancement] Added new test for using compression with parallel I/O:
  nc_test4/tst_h_par_compress.c. See
  [Unidata/netcdf-c#1784].

* [Bug Fix] Don't return error for extra calls to nc_redef() for netCDF/HDF5
  files, unless classic model is in use. See
  [Unidata/netcdf-c#1779].

* [Enhancement] Added new parallel I/O benchmark program to mimic NOAA UFS data
  writes, built when --enable-benchmarks is in configure. See
  [Unidata/netcdf-c#1777].

* [Bug Fix] Now allow szip to be used on variables with unlimited dimension
  [Unidata/netcdf-c#1774].

* [Enhancement] Add support for cloud storage using a variant of the Zarr
  storage format. Warning: this feature is highly experimental and is subject to
  rapid evolution
  [https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in].

* [Bug Fix] Fix nccopy to properly set default chunking parameters when not
  otherwise specified. This can significantly improve performance in selected
  cases. Note that if seeing slow performance with nccopy, then, as a
  work-around, specifically set the chunking
  parameters. [Unidata/netcdf-c#1763].

* [Bug Fix] Fix some protocol bugs/differences between the netcdf-c library and
  the OPeNDAP Hyrax server. Also cleanup checksum handling
  [https://github.com/Unidata/netcdf-c/issues/1712].* [Bug Fix] IMPORTANT: Ncgen
  was not properly handling large data sections. The problem manifests as
  incorrect ordering of data in the created file. Aside from examining the file
  with ncdump, the error can be detected by running ncgen with the -lc flag (to
  produce a C file). Examine the file to see if any variable is written in pieces
  as opposed to a single call to nc_put_vara. If multiple calls to nc_put_vara are
  used to write a variable, then it is probable that the data order is
  incorrect. Such multiple writes can occur for large variables and especially
  when one of the dimensions is unlimited.

* [Bug Fix] Add necessary __declspec declarations to allow compilation of netcdf
  library without causing errors or (_declspec related) warnings
  [Unidata/netcdf-c#1725].

* [Enhancement] When a filter is applied twice with different
parameters, then the second set is used for writing the dataset
[Unidata/netcdf-c#1713].

* [Bug Fix] Now larger cache settings are used for sequential HDF5 file creates/opens on parallel I/O capable builds; see [Github #1716](Unidata/netcdf-c#1716) for more information.

* [Bug Fix] Add functions to libdispatch/dnotnc4.c to support
   dispatch table operations that should work for any dispatch
   table, even if they do not do anything; functions such as
   nc_inq_var_filter [Unidata/netcdf-c#1693].

* [Bug Fix] Fixed a scalar annotation error when scalar == 0; see [Github
  #1707](Unidata/netcdf-c#1707) for more information.

* [Bug Fix] Use proper CURLOPT values for VERIFYHOST and VERIFYPEER; the
  semantics for VERIFYHOST in particular changed. Documented in NUG/DAP2.md. See
  [Unidata/netcdf-c#1684].

* [Bug Fix][cmake] Correct an issue with parallel filter test logic in
  CMake-based builds.

* [Bug Fix] Now allow nc_inq_var_deflate()/nc_inq_var_szip() to be called for
  all formats, not just HDF5. Non-HDF5 files return NC_NOERR and report no
  compression in use. This reverts behavior that was changed in the 4.7.4
  release. See [Unidata/netcdf-c#1691].

* [Bug Fix] Compiling on a big-endian machine exposes some missing forward
  delcarations in dfilter.c.

* [File Change] Change from HDF5 v1.6 format compatibility, back to v1.8
  compatibility, for newly created files.  The superblock changes from version 0
  back to version 2.  An exception is when using libhdf5 deprecated versions
  1.10.0 and 1.10.1, which can only create v1.6 compatible format.  Full
  backward read/write compatibility for netCDF-4 is maintained in all cases.
  See [Github #951](Unidata/netcdf-c#951).

## 4.7.4 - March 27, 2020

* [Windows] Bumped packaged HDF5 to 1.10.6, HDF4 to 4.2.14, and libcurl to
  7.60.0.

* [Enhancement] Support has been added for HDF5-1.12.0.  See
  [Unidata/netcdf-c#1528].

* [Bug Fix] Correct behavior for the command line utilities when directly
  accessing a directory using utf8 characters. See
  [Github #1669] (Unidata/netcdf-c#1669),
  [Github #1668] (Unidata/netcdf-c#1668) and
  [Github #1666] (Unidata/netcdf-c#1666) for more information.

* [Bug Fix] Attempts to set filters or chunked storage on scalar vars will now
  return NC_EINVAL. Scalar vars cannot be chunked, and only chunked vars can
  have filters. Previously the library ignored these attempts, and always
  storing scalars as contiguous storage. See
  [Unidata/netcdf-c#1644].

* [Enhancement] Support has been added for multiple filters per variable.  See
  [Unidata/netcdf-c#1584].

* [Enhancement] Now nc_inq_var_szip retuns 0 for parameter values if szip is not
  in use for var. See [Unidata/netcdf-c#1618].

* [Enhancement] Now allow parallel I/O with filters, for HDF5-1.10.3 and
  later. See [Unidata/netcdf-c#1473].

* [Enhancement] Increased default size of cache buffer to 16 MB, from 4
  MB. Increased number of slots to 4133. See
  [Unidata/netcdf-c#1541].

* [Enhancement] Allow zlib compression to be used with parallel I/O writes, if
  HDF5 version is 1.10.3 or greater. See
  [Unidata/netcdf-c#1580].

* [Enhancement] Restore use of szip compression when writing data (including
  writing in parallel if HDF5 version is 1.10.3 or greater). See
  [Unidata/netcdf-c#1546].

* [Enhancement] Enable use of compact storage option for small vars in
  netCDF/HDF5 files. See [Unidata/netcdf-c#1570].

* [Enhancement] Updated benchmarking program bm_file.c to better handle very
  large files. See [Unidata/netcdf-c#1555].

* [Enhancement] Added version number to dispatch table, and now check version
  with nc_def_user_format(). See
  [Unidata/netcdf-c#1599].

* [Bug Fix] Fixed user setting of MPI launcher for parallel I/O HDF5 test in
  h5_test. See [Unidata/netcdf-c#1626].

* [Bug Fix] Fixed problem of growing memory when netCDF-4 files were opened and
  closed. See [Unidata/netcdf-c#1575 and
  Unidata/netcdf-c#1571].

* [Enhancement] Increased size of maximum allowed name in HDF4 files to
  NC_MAX_NAME. See [Unidata/netcdf-c#1631].

## 4.7.3 - November 20, 2019

* [Bug Fix]Fixed an issue where installs from tarballs will not properly compile
  in parallel environments.

* [Bug Fix] Library was modified so that rewriting the same attribute happens
  without deleting the attribute, to avoid a limit on how many times this may be
  done in HDF5. This fix was thought to be in 4.6.2 but was not. See
  [Unidata/netcdf-c#350].

* [Enhancement] Add a dispatch version number to netcdf_meta.h and
  libnetcdf.settings, in case we decide to change dispatch table in future. See
  [Unidata/netcdf-c#1469].

* [Bug Fix] Now testing that endianness can only be set on atomic ints and
  floats. See [Unidata/netcdf-c#1479].

* [Bug Fix] Fix for subtle error involving var and unlimited dim of the same
  name, but unrelated, in netCDF-4. See
  [Unidata/netcdf-c#1496].

* [Enhancement] Update for attribute documentation. See
  [Unidata/netcdf-c#1512].

* [Bug Fix][Enhancement] Corrected assignment of anonymous (a.k.a. phony)
  dimensions in an HDF5 file. Now when a dataset uses multiple dimensions of the
  same size, netcdf assumes they are different dimensions. See [GitHub
  #1484](Unidata/netcdf-c#1484) for more information.

## 4.7.2 - October 22, 2019

* [Bug Fix][Enhancement] Various bug fixes and enhancements.

* [Bug Fix][Enhancement] Corrected an issue where protected memory was being
  written to with some pointer slight-of-hand.  This has been in the code for a
  while, but appears to be caught by the compiler on OSX, under circumstances
  yet to be completely nailed down.  See
  [GitHub #1486] (Unidata/netcdf-c#1486) for more information.

* [Enhancement] [Parallel IO] Added support for parallel functions in MSVC. See
  [Github #1492](Unidata/netcdf-c#1492) for more
  information.

* [Enhancement] Added a function for changing the ncid of an open file.  This
  function should only be used if you know what you are doing, and is meant to
  be used primarily with PIO integration. See
  [GitHub #1483] (Unidata/netcdf-c#1483) and
  [GitHub #1487] (Unidata/netcdf-c#1487) for more information.

## 4.7.1 - August 27, 2019

* [Enhancement] Added unit_test directory, which contains unit tests for the
  libdispatch and libsrc4 code (and any other directories that want to put unit
  tests there). Use --disable-unit-tests to run without unit tests (ex. for code
  coverage analysis). See
  [GitHub #1458] (Unidata/netcdf-c#1458).

* [Bug Fix] Remove obsolete _CRAYMPP and LOCKNUMREC macros from code. Also
  brought documentation up to date in man page. These macros were used in
  ancient times, before modern parallel I/O systems were developed. Programmers
  interested in parallel I/O should see nc_open_par() and nc_create_par(). See
  [GitHub #1459](Unidata/netcdf-c#1459).

* [Enhancement] Remove obsolete and deprecated functions nc_set_base_pe() and
  nc_inq_base_pe() from the dispatch table. (Both functions are still supported
  in the library, this is an internal change only.) See [GitHub
  #1468](Unidata/netcdf-c#1468).

* [Bug Fix] Reverted nccopy behavior so that if no -c parameters are given, then
  any default chunking is left to the netcdf-c library to decide. See [GitHub
  #1436](Unidata/netcdf-c#1436).

## 4.7.0 - April 29, 2019

* [Enhancement] Updated behavior of `pkgconfig` and `nc-config` to allow the use
  of the `--static` flags, e.g. `nc-config --libs --static`, which will show
  information for linking against `libnetcdf` statically. See
  [Github #1360] (Unidata/netcdf-c#1360) and
  [Github #1257] (Unidata/netcdf-c#1257) for more information.

* [Enhancement] Provide byte-range reading of remote datasets. This allows
  read-only access to, for example, Amazon S3 objects and also Thredds Server
  datasets via the HTTPService access method.  See
  [GitHub #1251] (Unidata/netcdf-c#1251).

* Update the license from the home-brewed NetCDF license to the standard
  3-Clause BSD License.  This change does not result in any new restrictions; it
  is merely the adoption of a standard, well-known and well-understood license
  in place of the historic NetCDF license written at Unidata.  This is part of a
  broader push by Unidata to adopt modern, standardized licensing.

## 4.6.3 - February 28, 2019

* [Bug Fix] Correctly generated `netcdf.pc` generated either by `configure` or
  `cmake`.  If linking against a static netcdf, you would need to pass the
  `--static` argument to `pkg-config` in order to list all of the downstream
  dependencies.  See
  [Github #1324](Unidata/netcdf-c#1324) for more information.

* Now always write hidden coordinates attribute, which allows faster file opens
  when present. See
  [Github #1262](Unidata/netcdf-c#1262) for more information.

* Some fixes for rename, including fix for renumbering of varids after a rename
  (#1307), renaming var to dim without coordinate var. See
  [Github #1297] (Unidata/netcdf-c#1297).

* Fix of NULL parameter causing segfaults in put_vars functions. See
   [Github #1265] (Unidata/netcdf-c#1265) for more information.

* Fix of --enable-benchmark benchmark tests
   [Github #1211] (Unidata/netcdf-c#1211)

* Update the license from the home-brewed NetCDF license to the standard
  3-Clause BSD License.  This change does not result in any new restrictions; it
  is merely the adoption of a standard, well-known and well-understood license
  in place of the historic NetCDF license written at Unidata.  This is part of a
  broader push by Unidata to adopt modern, standardized licensing.

* [BugFix] Corrected DAP-related issues on big-endian machines. See
  [Github #1321] (Unidata/netcdf-c#1321),
  [Github #1302] (Unidata/netcdf-c#1302) for more information.

* [BugFix][Enhancement] Various and sundry bugfixes and performance
  enhancements, thanks to \@edhartnett, \@gsjarrdema, \@t-b, \@wkliao, and all
  of our other contributors.

* [Enhancement] Extended `nccopy -F` syntax to support multiple variables with a
  single invocation. See
  [Github #1311](Unidata/netcdf-c#1311) for more information.

* [BugFix] Corrected an issue where DAP2 was incorrectly converting signed
  bytes, resulting in an erroneous error message under some circumstances. See
  [GitHub #1317] (Unidata/netcdf-c#1317) for more
  information.  See
  [Github #1319] (Unidata/netcdf-c#1319) for related information.

* [BugFix][Enhancement] Modified `nccopy` so that `_NCProperties` is not copied
  over verbatim but is instead generated based on the version of `libnetcdf`
  used when copying the file.  Additionally, `_NCProperties` are displayed
  if/when associated with a netcdf3 file, now. See
  [GitHub#803] (Unidata/netcdf-c#803) for more information.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants