The failure comes at the end when each processor tries to
remove() the created file. The second one appears to fail
-- possibly because of a race condition. In any case, if I
move the remove() call past MPI_Finalize(), the error goes away.
It is not clear of the exact circumstances that cause the
failure, but I encountered it with the following env.
- automake on centos7 64-bit.
- compiling with -O1 for gcc.
- The pnetcdf version is 1.8.1
- The command "mpiexec --version" gives:
Version: 3.0.4
Release Date: Wed Apr 24 10:08:10CDT2013
re pull request https://github.com/Unidata/netcdf-c/pull/405
re pull request https://github.com/Unidata/netcdf-c/pull/446
Notes:
1. This branch is a cleanup of the magic.dmh branch.
2. magic.dmh was originally merged, but caused problems with parallel IO.
It was re-issued as pull request https://github.com/Unidata/netcdf-c/pull/446.
3. This branch + pull request replace any previous pull requests and magic.dmh branch.
Given an otherwise valid netCDF file that has a corrupted header,
the netcdf library currently crashes. Instead, it should return
NC_ENOTNC.
Additionally, the NC_check_file_type code does not do the
forward search required by hdf5 files. It currently only looks
at file position 0 instead of 512, 1024, 2048,... Also, it turns
out that the HDF4 magic number is assumed to always be at the
beginning of the file (unlike HDF5).
The change is localized to libdispatch/dfile.c See
https://support.hdfgroup.org/release4/doc/DSpec_html/DS.pdf
Also, it turns out that the code in NC_check_file_type is duplicated
(mostly) in the function libsrc4/nc4file.c#nc_check_for_hdf.
This branch does the following.
1. Make NC_check_file_type return NC_ENOTNC instead of crashing.
2. Remove nc_check_for_hdf and centralize all file format checking
NC_check_file_type.
3. Add proper forward search for HDF5 files (but not HDF4 files)
to look for the magic number at offsets of 0, 512, 1024...
4. Add test tst_hdf5_offset.sh. This tests that hdf5 files with
an offset are properly recognized. It does so by prefixing
a legal file with some number of zero bytes: 512, 1024, etc.
5. Off-topic: Added -N flag to ncdump to force a specific output dataset name.
The nc_test/util.c error is a typo. The libdap4/d4meta.c error only is shown
when using a 64 bit machine because then |size_t| == 64 bits
and |int| = 32bits.
1. When running under windows (as opposed to cygwin)
we need to make sure to not user /cygdrive/ file paths.
This was ocurring in libdap4/d4read.c, but may occur
elsewhere.
2. Shell scripts in the git repo are not being checked-out
with the executable mode set. Had core.filemode set to false.
Was a major hassle to fix.
Specific changes:
1. Add dap4 code: libdap4 and dap4_test.
Note that until the d4ts server problem is solved, dap4 is turned off.
2. Modify various files to support dap4 flags:
configure.ac, Makefile.am, CMakeLists.txt, etc.
3. Add nc_test/test_common.sh. This centralizes
the handling of the locations of various
things in the build tree: e.g. where is
ncgen.exe located. See nc_test/test_common.sh
for details.
4. Modify .sh files to use test_common.sh
5. Obsolete separate oc2 by moving it to be part of
netcdf-c. This means replacing code with netcdf-c
equivalents.
5. Add --with-testserver to configure.ac to allow
override of the servers to be used for --enable-dap-remote-tests.
6. There were multiple versions of nctypealignment code. Try to
centralize in libdispatch/doffset.c and include/ncoffsets.h
7. Add a unit test for the ncuri code because of its complexity.
8. Move the findserver code out of libdispatch and into
a separate, self contained program in ncdap_test and dap4_test.
9. Move the dispatch header files (nc{3,4}dispatch.h) to
.../include because they are now shared by modules.
10. Revamp the handling of TOPSRCDIR and TOPBUILDDIR for shell scripts.
11. Make use of MREMAP if available
12. Misc. minor changes e.g.
- #include <config.h> -> #include "config.h"
- Add some no-install headers to /include
- extern -> EXTERNL and vice versa as needed
- misc header cleanup
- clean up checking for misc. unix vs microsoft functions
13. Change copyright decls in some files to point to LICENSE file.
14. Add notes to RELEASENOTES.md
Update utf8proc.[ch] to use the version now
maintained by the Julia Language project
(https://github.com/JuliaLang/utf8proc/blob/master/LICENSE.md).
The license for the previous version was
unacceptable for the Debian and Ubuntu release
systems. The new version both updates the code
and addresses the license issue.
It turns out that the utf8proc software we are using
was turned over to the Julia Language developers
and the license terms changed to allow modification.
(https://github.com/JuliaLang/utf8proc/blob/master/LICENSE.md).
So the fix here is as follows:
1. Wrap the library with a fixed interface: libdispatch/dutf8.c
and include/ncutf8.h.
2. Replace the existing utf8proc code with the new version
from https://github.com/JuliaLang/utf8proc.
3. Add a couple more test cases: nc_test/tst_utf8_validate.c
and nc_test_utf8_phrases.c. If/when I can find a usable
normalization test, I will incorporate that later.
This consists of a persistent attribute named
_NCProperties plus two computed attributes
_IsNetcdf4 and _SuperblockVersion.
See the 'Provenance Attributes' section
of docs/attribute_conventions.md for details.
AC_CHECK_SIZEOF is not working because anti-virus
will not allow very rapid creation/deletion of a
file with same name.
2. modified some test baselines to attempt to fix
Ward's issue
AC_CHECK_SIZEOF is not working because anti-virus
will not allow very rapid creation/deletion of a
file with same name.
2. modified some test baselines to attempt to fix
Ward's issue
This supports better authorization
handling for DAP requests, especially redirection
based authorization. I also added a new test case
ncdap_tests/testauth.sh.
Specifically, suppose I have a netrc file /tmp/netrc
containing this.
machine uat.urs.earthdata.nasa.gov login xxxxxx password yyyyyy
Also suppose I have a .ocrc file containing these lines
HTTP.COOKIEJAR=/tmp/cookies
HTTP.NETRC=/tmp/netrc
Assume that .ocrc is in the local directory or HOME.
Then this command should work (assuming a valid login and password).
ncdump -h "https://54.86.135.31/opendap/data/nc/fnoc1.nc"
m4 can emit 'sync lines' which tell the toolchain "hey, this code
actually came from this other file over here". Should help prevent
anyone accidentally editing a generated file.
The pnetcdf support was not
properly being used to provide
mpi parallel io for netcdf-3 classic
files. The wrong dispatch table was being
used. The fix was to modify
dfile.c#NC_check_file_type to properly
specify the pnetcdf dispatch table when
use_parallel was true.
extract info from libnetcdf.settings
API is below.
I have made this API public yet
by adding it to netcdf.h. I will
do that when everyone is agreed on the
proper API.
extern const char* nc_settings(const char* key); /*get value of a specific key */
extern const char** nc_settings_all(); /*get all settings in envv format */
extern void nc_settings_reclaim(); /* reclaim all space and clean up */
Envv format is
{key,value}*,NULL
Also added test: nc_test/tst_settings.c
Add a new function called nc_inq_format_extended that
returns more detailed format information (vis-a-vis
nc_inq_format) about an open dataset.
Note that the netcdf API will present the file as if it had
the format specified by nc_inq_format. The true file
format, however, may not even be a netcdf file; it might be
DAP, HDF4, or PNETCDF, for example. This function returns
that true file type. It also returns the effective mode for
the file.
signature: nc_inq_format_extended(int ncid, int* formatp, int* modep)
where
* ncid is the NetCDF ID from a previous call to nc_open() or
nc_create().
* formatp is a pointer to a location for returned true format.
* modep is a pointer to a location for returned mode flags.
Refer to the actual list in the file netcdf.h to see the
currently defined set.
Also added test cases (tst_formatx*).
to clean up resources properly on failure.
Refactored doubly-linked list code for objects in the libsrc4 directory,
cleaning up the add/del routines, breaking out the common next/prev
pointers into a struct and extracting the add/del operations on them,
changed the list of dims to add new dims in the same order as the other
types, made all add routines able to optionally return a pointer to the
newly created object.
Removed some dead code (pg_var(), nc4_pg_var1(), nc4_pg_varm(), misc. small
routines, etc)
Fixed fill value handling for string types in nc4_get_vara().
Changed many malloc()+strcpy() pairs into calls to strdup().
Cleaned up misc. other minor Coverity issues.
Add support for embedded NUL characters
in character arrays and added some test cases.
Note that embedded NULs in netcdf-4
string constants are still illegal. This is
because strings are not counted in the netcdf
API, so they are implicitly NUL terminated.
netCDF classic or 64-bit offset files that have a UINT32_MAX flag for
large last record size of a variable that has values larger than 1
byte. This problem had previously been fixed for *writing* such data,
but was only tested with an ncbyte variable. Fixed test to
demonstrate problem and the fix.
More updates to chunking documentation, cosmetic fixes for some
"--option=" documentation that doxygen turns into mdash.
revisions are simpler, and I hope, clearer than the all
previous versions. Part of the code taken from the
ucar.nc2.Range.compose function. Modified to handle edge
cases not handled in the Range.compose function.
enhancements, based on contributed code from Martin van Driel, to
support -v, -g, -V, and -G options for selecting groups and variables
in output. Fix all clang warnings from nccopy and ncdump sources, as
well as a few other cleanup changes to testing code.
CMake related changes in CMakeLists.txt files,
cmake_config.h.in. Other changes relate to
Windows-specific issues, and changes made
when regenerating generated source files.
reported by static analysis, including memory leak in ncdump, missing
size_t cast for chunk cache. Fixed various doc problems, including
byte vs. char issues, missing NC_UBYTE in type list, needed link to
"Building with Windows" page.
implementation. Deleted obsolete win32, soon to be replaced by Ward's
Windows 32- and 64-bit fixes for building with MSYS/MinGW. Made
cosmetic cleanup to output of "make check" to make it easier for users
to interpret. Fixed bug NCF-175: ncdump -t incorrectly interpreting
units attribute (such as "days") without a base time (such as "since
2007-01-01") as a time unit.
Changed name to 4.2.1-beta.
of NC_ERANGE errors, especially with unsigned types, resolving most
NCF-172 issues and confusion between long and long long types in
llibsrc/ncx.m4. Eliminated some unnecessary tests, e.g. tests
for negative values for unsigned types.
in code and scripts.
Fixed bug NCF-171: error reading external int into longlong or writing
from longlong array to external int on 32-bit platforms and classic
format files.
Promoted test demonstrating NCF-171 bug from longlong array from
"extra test" to "test".
range_error checks in netCDF-4 type conversion code. Made netCDF
attribute tests with type conversion more comprehensive and stringent,
fixing bugs identified with better tests. Changed a test in
nc_test/tst_atts.c to use netCDF-3 file instead of netCDF-4 file,
because that directory is supposed to be for tests that work with
--disable-netcdf-4. Added test demonstrating NCF-171 bug on 32-bit
platforms, only run when configured with --enable-extra-tests.
The in-memory files can be made persistent if nc_create is called with
NC_DISKLESS|NC_WRITE flags set. Initial test case also included.
- Modified ncio mechanism to support
multiple ncio packages; this is so we
can have posixio and memio operating
at the same time.
- cleanup up a bunch of lint issues (unused variables, etc).
