Open importmattie opened 8 years ago
Works on my mac. Looks like a segfault in the C lib - perhaps a bug that has been fixed? What version of the netcdf c lib are you using? You can find out using:
>>> import netCDF4
>>> netCDF4.__version__
'1.2.2'
>>> netCDF4.__hdf5libversion__
u'1.8.15-patch1'
>>> netCDF4.__netcdf4libversion__
u'4.4.0-development'
>>> netCDF4.Dataset('https://weather.rsmas.miami.edu/repository/opendap/synth:eab82de2- d682-4dc0-ba8b-2fac7746d269:L2FsbFZhcnNfcjkweDQ1XzMubmM0/entry.das','r')
<type 'netCDF4._netCDF4.Dataset'>
root group (NETCDF3_CLASSIC data model, file format DAP2):
CDI: Climate Data Interface version 1.6.4 (http://code.zmaw.de/projects/cdi)
history: Thu Oct 29 09:47:06 2015: cdo remap,../Weights/r90x45_overlap_source.txt,../Weights /r90x45_overlap_from2.nc4 allVars_r180x90_1_noHMV.nc4 allVars_r90x45_1.nc4
Thu Oct 29 09:19:11 2015: ncks -x -v HMV allVars_r180x90_1.nc4 allVars_r180x90_1_noHMV.nc4
Wed Oct 28 15:20:51 2015: ncks -x -v KEDOT allVars_r180x90_1.nc4 allVars_r180x90_1_noKEDOT.nc4
Wed Oct 28 15:16:12 2015: ncatted -a units,SKEDOT_ZON,o,c,m2 s-3 /home/niznik/niznik- data2/shared_repository/allVars_r180x90_1.nc4
Wed Oct 28 15:16:12 2015: ncatted -a long_name,SKEDOT_ZON,o,c,eddy vertical momentum flux /home/niznik/niznik-data2/shared_repository/allVars_r180x90_1.nc4
Wed Oct 28 15:13:39 2015: ncks -A -v SKEDOT_ZON /home/niznik/niznik-data2/calcVars_3D /SKEDot_r180x90.nc4 /home/niznik/niznik-data2/shared_repository/allVars_r180x90_1.nc4
Thu Jun 4 14:33:00 2015: ncatted -a long_name,KEDOT,o,c,eddy vertical momentum flux allVars_r180x90_1.nc4
Thu Jun 4 14:26:07 2015: ncks -x -v KEDot allVars_r180x90_1.nc4 allVars_r180x90_1_fix.nc4
Thu Jun 4 14:22:08 2015: ncks -A -v KEDOT /home/niznik/niznik-data2/calcVars_3D /KEDot_r180x90.nc4 /home/niznik/niznik-data2/allVars_r180x90_1.nc4
Thu Jun 04 00:22:57 2015: cdo remap,r180x90,Weights/r180x90_fromP5.nc4 calcVars.nc4 allVars_r180x90_1.nc4
Conventions: CF-1.4
NCO: 4.3.7
CDO: Climate Data Operators version 1.6.4 (http://code.zmaw.de/projects/cdo)
nco_openmp_thread_number: 1
DODS_EXTRA.Unlimited_Dimension: time
dimensions(sizes): time(6087), lat(45), lon(90)
variables(dimensions): float64 lon(lon), float64 lat(lat), float64 time(time), float32 PREC(time,lat,lon), float32 Q(time,lat,lon), float32 SKEDOT_ZON(time,lat,lon), float32 T(time,lat,lon), float32 TEF(time,lat,lon), float32 U(time,lat,lon), float32 UU(time,lat,lon), float32 V(time,lat,lon), float32 VV(time,lat,lon), float32 W(time,lat,lon), float32 WPQP(time,lat,lon), float32 WPTP(time,lat,lon), float32 WPUP(time,lat,lon), float32 WQ(time,lat,lon), float32 WT(time,lat,lon), float32 WU(time,lat,lon), float32 WV(time,lat,lon), float32 HMV(time,lat,lon)
groups:
Yeah, it does look like a few of them are out of date. Here's the output, and in the meantime I'll try updating the other libraries and see if that fixes things.
In [6]: netCDF4.version
Out[6]: '1.1.9'
In [7]: netCDF4.hdf5libversion
Out[7]: u'1.8.15-patch1'
In [8]: netCDF4.netcdf4libversion
Out[8]: u'4.3.3.1'
I just tried here with netcdf 4.3.3.1 and hdf5 1.8.15-patch1, and it works.
Just to note, downgrading to 1.1.8 worked for me as a temporary fix. I will try 1.2.2 soon.
I use anaconda python on a MacBook Pro running OS X El Capitan. I updated my python packages via anaconda this morning and then re-added netCDF4 (1.1.9-np110py27_0).
Afterwards, when I call netCDF4.Dataset to load remote data I get a segmentation fault. Here's the simplest code that will lead to the segmentation fault:
netCDF4.Dataset('https://weather.rsmas.miami.edu/repository/opendap/synth:eab82de2-d682-4dc0-ba8b-2fac7746d269:L2FsbFZhcnNfcjkweDQ1XzMubmM0/entry.das','r')
I did not get the same error when using that call on a file on the local hard drive. If it helps, I've also attached the Mac error report generated when this happens in an IPython Notebook.
errorReport.txt